article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
the fate of the most massive stars in galaxies is not well understood .
such stars have been theoretically linked to black - hole formation , to some gamma - ray burst sources , and to early metal enrichment in the universe ( e.g. , umeda & nomoto 2003 ) . in principle , stars with initial masses @xmath4@xmath5 should pass through the red supergiant phase , or directly to the short - lived luminous blue variable ( lbv ) phase , on to the wolf - rayet ( wr ) phase ( e.g. , langer et al .
1994 ; stothers & chin 1996 ) , before exploding as supernovae ( sne ; e.g. , woosley , langer , & weaver 1993 ) or ending as `` collapsars '' ( e.g. , macfadyen & woosley 1999 ) . the great eruption of 1843 for @xmath0 carinae ( see davidson & humphreys 1997 ) , where the star s bolometric luminosity increased by a very large amount , demonstrated that some very massive stars go through spectacular eruptive phases of pre - sn mass ejection . this instability may play a role in shaping the appearance of the upper hertzsprung - russell ( h - r ) diagram ( humphreys & davidson 1994 ) .
however , lbvs , and particularly phenomenal cases such as @xmath0 car , are extremely rare ( humphreys & davidson 1994 ) ; isolating additional examples is therefore necessary for a better understanding of the evolution of very massive stars . the identification of a number of extragalactic objects as genuine sne ( strictly defined to be the violent destruction of a star at the end of its life ) has been questioned by several teams of investigators ( e.g. , goodrich et al
. 1989 ; filippenko et al . 1995 ; van dyk et al . 2000 , 2004 ; smith , humphreys , & gehrz 2001 ; wagner et al . 2004 ) .
these authors have argued that the `` sn impostors '' are more likely the pre - sn cataclysmic event of a very massive star , analogous to @xmath0 car ( hence , the term @xmath0 car analogs " or `` @xmath0 car variables '' ; humphreys , davidson , & smith 1999 ) .
similarly to @xmath0 car , the precursor star is expected to survive the eruptive super - outburst .
the prototypical examples are sn 1961v in ngc 1058 ( zwicky 1964 ; goodrich et al . 1989 ; filippenko et al .
1995 ; humphreys et al . 1999 ; van dyk , filippenko , & li 2002 ) and sn 1954j in ngc 2403 ( humphreys et al . 1999 ; smith et al .
five more - recent examples are sn 1997bs in m66 ( van dyk et al .
2000 ) , sn 1999bw in ngc 3198 ( filippenko , li , & modjaz 1999 ) , sn 2000ch in ngc 3432 ( filippenko 2000 ; wagner et al . 2004 ) , sn 2001ac in ngc 3504 ( matheson et al . 2001 ) , and sn 2002 kg , also in ngc 2403 ( schwartz & li 2002 ) .
the star v1 in ngc 2366 ( drissen , roy , & robert 1997 ; drissen et al .
2001 ) may be an additional member of this group .
for sn 1997bs ( van dyk et al .
1999 ) and sn 2002 kg ( weis & bomans 2005 ; van dyk et al .
2005 ) , luminous , very massive precursor stars have been directly identified . whereas the early - time optical spectrum of sn 1961v was dominated by narrow emission lines of h , he i , and fe ii ( indicating a relatively low maximum expansion velocity of @xmath6 km s@xmath7 ; zwicky 1964 ) , similar to what is seen for
the type iin " sn classification ( schlegel 1990 ; filippenko 1997 ) , no spectra of sn 1954j exist at a comparable epoch .
the identification of this latter event as a sn , therefore , has always been debatable .
kowal et al .
( 1972 ) considered it a sn , possibly of type v " ( zwicky 1965 ) , based on the relative similarity of its light curve to the highly peculiar one of sn 1961v .
tammann & sandage ( 1968 ) had identified the object as an irregular blue variable , variable 12 ( v12 ; see their figure 11 for the unusual light curve ) . in quiescence
the star showed only minor variability , with @xmath8 mag , but subsequently in outburst it fluctuated drastically in brightness over a 5-yr span , beginning in 1949 , reaching a maximum @xmath9 mag in mid - to - late 1954 , when it was the brightest star in ngc 2403 . by late 1954 , it had declined steeply to a level fainter than its pre - outburst state .
( sn 1954j is still seen brightly in the poss - i plates , taken on 1955 january 29 ut ! )
humphreys et al .
( 1999 ) considered this behavior for sn 1954j / v12 quite unlike what is seen for an actual sn .
no survivor of an extragalactic super - outburst has yet been unambiguously recovered years to decades afterwards .
the most contentious case is sn 1961v ( see chu et al .
2004 ; van dyk 2005 , and references therein ) , where an @xmath0 car - like star has been detected in the sn environment , but it is still uncertain if it can be directly linked to a putative survivor . a more clear - cut case may be sn 1954j / v12 , where smith et al .
( 2001 ) first identified , through ground - based optical ( @xmath10 ) and near - infrared ( @xmath11 ) imaging , an object they considered the post - outburst survivor star . here
we contribute further imaging of sn 1954j / v12 from the ground with the keck - i 10-m telescope and the 2.56-m nordic optical telescope , confirming the source identified by smith et al .
more importantly , we demonstrate with the superior spatial resolution of the _ hubble space telescope _ ( _ hst _ ) that the object seen from the ground is actually a group of stars .
furthermore , we are able to isolate a star within the group as the likely candidate for the 1954 super - outburst survivor . by combining these data with an optical spectrum from the keck - ii telescope
, we argue that this star is a rare , very luminous and massive , @xmath0 car - like variable star .
we assume throughout this paper that the true distance modulus to ngc 2403 is @xmath12 mag ( freedman et al .
2001 ) . at this distance
, 1 corresponds to 15.2 pc .
sn 1954j / v12 was imaged in @xmath13 with the low resolution imaging spectrometer ( lris ; oke et al .
1995 ) on the keck - i telescope on 1996 october 20 ( ut dates are used throughout this paper ) .
exposure times were 240 s in each band , and the seeing was quite good in the @xmath14-band ( @xmath15 ) , less so at @xmath16 ( @xmath17 ) , and mediocre at @xmath18 ( @xmath19 , largely due to less - than - optimal focus ) .
standard image - calibration procedures ( bias subtraction , flat - fielding , etc . ) were employed in the reduction of these data .
we extracted photometry from these images using iraf / daophot ( stetson 1987 ) and fitting an empirically derived model point - spread function ( psf ) for each band .
unfortunately , no photometric calibration was obtained during this night ; thus , we had to tie the photometry to the calibrated photometry obtained from the nordic optical telescope ( not ) images ( see 2.2 ) .
from the digitized red poss - i image ( the digitized sky survey ; dss ) we measure an accurate absolute position for sn 1954j / v12 of @xmath20(j2000 ) = @xmath21 , @xmath22(j2000 ) = @xmath23 , with a total uncertainty of @xmath24 , using the two micron all sky survey ( 2mass ) as the astrometric reference ( see van dyk , li , & filippenko 2003 ) . applying the same astrometric grid to the keck images , we confirm in these images at this position the source identified by smith et al .
( 2001 ; see their figure 2 ) . in figure 1 ( left panel )
we show the @xmath14-band image , revealing that the source appears extended with an `` appendage '' @xmath25 to the southwest .
the appearance of the source changes in the @xmath16 band ; see figure 1 ( right panel ) .
two closely spaced sources are identified by daophot in this band , with the `` appendage '' source becoming more obvious , and the main source to the north containing less light in this band .
we identify the main source , seen clearly in the left panel , as the smith et al .
( 2001 ) object and list its photometry from the keck images in table 1 , along with the photometry from smith et al .
( 2001 ) .
larsen & richtler ( 1999 ) obtained @xmath10 ( plus h@xmath20 ) images of ngc 2403 with the 2.56-m not on 1997 october 13 , under very good seeing conditions of @xmath26 .
these images were contributed to the nasa / ipac extragalactic database ( ned ) and posted for public distribution on their website .
we extracted photometry from the broad - band images , first with a 4 aperture and then using psf fitting within iraf / daophot .
the resulting instrumental psf - fitting magnitudes , after aperture corrections , were calibrated using photometry of a number of isolated stars in the images through a 20 aperture , which matches the aperture used to establish the calibration through observations of standard stars by larsen ( 1999 ; his table 1 and 3.1 ; we also first applied the appropriate extinction corrections , supplied via a private communication from s. s. larsen ) .
we have compared the overall photometry with the @xmath27 bright - star photometry in sandage ( 1984 ) and @xmath28 photometry in zickgraf & humphreys ( 1991 ) and find very good agreement across all bands .
the same astrometric grid applied to the dss and keck images was also applied to the not images , again confirming the smith et al .
( 2001 ) source .
the varying shape of the source as a function of wavelength , as seen in the keck images , is also seen in the not data . in particular , the `` appendage '' is also detected in the not @xmath16-band image .
the @xmath10 photometry for the source from the not images is given in table 1 .
an optical spectrum of the sn 1954j / v12 site was obtained with the echellette spectrograph and imager ( esi ; sheinis et al .
2002 ) on the keck - ii telescope on 2002 november 9.56 ( exposure time 2400 s ) .
the 1 slit included both the smith et al .
( 2001 ) source and the `` appendage , '' but the seeing was poor ( @xmath29 ) and the two sources were blended .
variable seeing ( it was @xmath30 during the observation of the standard star ) and the presence of thin cirrus compromised the absolute flux calibration to some degree .
conventional data reductions were performed within iraf , including bias subtraction , flat - fielding , and wavelength calibration , while our own idl routines were used for flux calibration and removal of telluric lines ( e.g. , matheson et al .
the extraction from the relevant order containing h@xmath20 ( the only feature detected in the overall spectrum ) is shown in figure 2 .
although the spectrum is somewhat noisy , broad - line emission with full - width at half - maximum ( fwhm ) of @xmath311650 km s@xmath7 is clearly seen .
the gaussian velocity width , @xmath32 , of the line is then @xmath31700 km s@xmath7 . the equivalent width of this emission line is @xmath31240 .
we suspect the origin of this emission is either within the extended source or from the appendage , but can not tell from the spectrum alone ( we discuss this in more detail below ) .
the recent , bright sn 2004dj in ngc 2403 was imaged by program go-10182 ( pi : filippenko ) with the advanced camera for surveys ( acs ) on - board _ hst _ on 2004 august 17 .
both the wide field channel ( wfc ) and the high resolution channel ( hrc ) were used , and the resulting data immediately became publicly available in the _ hst _ archive .
images were obtained with the wfc in the bands f475w , f606w , f814w , and f658n , with respective exposure times of 600 , 350 , 350 , and 650 s. the imaging strategy for sn 2004dj was purposely designed so that the site of sn 1954j / v12 would also be contained in the wfc images .
we recombined the individual flat - fielded `` flt '' exposures in all four bands at the space telescope science institute ( stsci ) , in order to more reliably remove cosmic - ray hit features from the images , and , in effect , produce our own drizzled ( fruchter & hook 2002 ) `` drz '' images independent of the _ hst _ data pipeline . in figure 3
we show the f606w , f814w , and f658n images .
comparison with figure 1 ( in particular , see the inset to figure 3a ) clearly shows that the source identified from the ground is actually a resolved group ( or cluster ) of stars .
we label in figure 3 the four brightest members of the cluster , in decreasing order of their f606w apparent brightness . since the stellar environment is crowded , we measured the brightnesses of the four stars in the three broad - band images and the one narrow - band image using daophot , fitting the stellar profiles with psfs , based on a @xmath33 aperture and generated from the @xmath34 brightest , uncrowded stars in a @xmath35 pixel area around the sn site .
we verified that this produces , to within the measurement uncertainties , equivalent results as measuring , first , the instrumental magnitudes through a @xmath36 aperture and , second , correcting these magnitudes to those through a @xmath33 aperture , at least for the two blue broad bands ( aperture corrections for the f814w band are more complex , due to the possible presence of the red halo feature around stars ; see sirianni et al .
we determined that any correction to the magnitudes for the degradation of charge - transfer efficiency ( cte ) is @xmath370.6% in the three broad bands ( see riess & mack 2004 ) .
the cte correction for the f658n band measurements , however , was more appreciable , at @xmath312% . for the f475w , f606w , and f658n measurements we next applied the correction from a @xmath33 aperture to infinite aperture from table 5 in sirianni et al .
( 2005 ) ; the same is also true for the blue star 3 in the f814w band . for stars 1 , 2 , and 4 in the f814w band we applied the corrections from figure 12 in sirianni
we then applied the zeropoints in table 11 of sirianni et al . to the corrected instrumental magnitudes , vegamag for the broad bands and stmag for the narrow band
the resulting flight system magnitudes are given in table 2 .
we also transform the broad - band magnitudes into the standard johnson - cousins system , following the translation established by sirianni et al .
, with the understanding that this transformation is fraught with uncertainties , especially for objects with non - standard photospheres .
furthermore , we do not know _ a priori _ the extinction toward the four stars , so we perform the transformation for each star first , then establish the extinction , contrary to the recommendation by sirianni et al .
in table 2 we list the resulting @xmath38 magnitudes for the four stars .
we suggest that what is seen from the ground for sn 1954j / v12 is the composite of the stars resolved in the _ hst _ images , with the various individual stars contributing different amounts to the overall brightness at different wavelengths .
this is borne out in table 1 , where at @xmath38 we sum up the flux contributions from stars 1 , 3 , and 4 .
( when the wfc images are smoothed using a gaussian with widths corresponding to the ground - based seeing for the keck and not images , it is readily apparent that the smith et al .
2001 source consists of these three stars . )
star 2 is almost certainly the `` appendage '' to the main source , as seen from the ground , and we therefore do not include the contribution from its light .
these sums at each band agree quite well with the brightness of sn 1954j / v12 in the ground - based measurements .
however , we note that the @xmath39 and @xmath16 magnitudes from smith et al .
( 2001 ) are generally fainter than the other measurements .
( smith et al . employed a 1 aperture with 2 seeing for their measurements , and for the @xmath16 band the measurement
is tied to one comparison star , without an adequate set of observed standard stars , which could account for some of the discrepancies . )
also , our @xmath40 measurement from the 1997 image data and our @xmath18 measurement from the 1996 data are somewhat brighter than the ensemble in that band , but we have already noted the degraded focus in the keck @xmath18-band image .
we can not exclude that one or more of the stars in the environment , even sn 1954j / v12 itself , is somewhat variable .
however , it appears that the brightness and color of the smith et al .
( 2001 ) source have been relatively constant overall between 1996 and 2004 .
the properties ( line width , equivalent width ) of the broad h@xmath20 emission line in sn 1954j / v12 , as seen in the ground - based spectrum , are strikingly similar to those of @xmath0 car itself . in figure 4
we directly compare the sn 1954j / v12 spectrum with a recent ( 1998 january 1 ) spectrum of @xmath0 car obtained with _
hst_/stis by the _ hst _
treasury program ( go-9420 ; pi : k. davidson ) on @xmath0 car .
the broad components of the two spectra agree quite well , although the narrower component , as well as the pronounced p - cygni profile in the broader component , seen in the @xmath0 car spectrum are both noticeably absent in the sn 1954j / v12 spectrum .
we contend that this late - time spectrum of sn 1954j / v12 is not of a decades - old sn : the spectrum of sn 1957d in m83 , of comparable age , is dominated by lines due to oxygen with velocities in excess of @xmath312000 km s@xmath7 ( long , winkler , & blair 1992 ) .
we also consider the probability of such a rare @xmath0 car - like object being in the immediate vicinity of sn 1954j / v12 ( which photometrically resembles @xmath0 car ; humphreys et al .
1999 ) , yet _ not _ being the sn itself , to be extremely small .
it is most likely that one of the four brightest stars at the sn site seen in the _ hst _ images is the source of the broad h@xmath20 emission in the keck spectrum , and that it is the survivor of the sn 1954j / v12 outburst .
but , which one ? examining figure 3c we see that , within the environment , one star has a clear , luminous excess of emission in the f658n band , relative to the other three stars . comparing with figures 3a and 3b , star 4 is exactly at the position of this emission source . from table 2
we see that the stmag for star 4 at f658n is @xmath41 mag ( without any continuum subtraction ) , which corresponds to an equivalent flux density per unit wavelength of @xmath42 erg s@xmath7 @xmath43 @xmath7 .
we note that this agrees quite well with the value of @xmath44 near 6580 in the spectrum shown in figure 2 .
( this agreement assumes that the line flux has not changed between 2002 november and 2004 august , and also may be fortuitous , given the variable seeing conditions and thin cirrus present at the time of the keck observations . )
thus , we find it suggestive that the source of the broad h@xmath20 emission is indeed star 4 .
this is further supported by inspection of the 2-dimensional esi spectrum of the specific order ( # 9 ) containing h@xmath20 . in figure
1a we label a star ( `` a '' ) , to the northeast of the sn environment , which also fell within the esi slit . using this star as a fiducial , and measuring the spatial offset between the spatial peak of this fiducial star s spectrum and that of the h@xmath20 source
, we find @xmath45 . measuring the offsets between this star and stars in the grouping on the acs / wfc images
, we find @xmath46 for star 1 , @xmath47 for star 2 , @xmath48 for star 3 , and @xmath49 for star 4 ( with uncertainties of @xmath50 wfc pixel , i.e. , @xmath51 ) . we can reject stars 1 and 2 as the h@xmath20 source , given their large displacement from the fiducial .
star 3 , at face value , is in closest agreement with the spectrum offset ; however , we see very weak emission in the f658n image at its position ( see table 2 ) .
this , again , leaves star 4 as the most likely candidate for the emission seen in the spectrum , whose displacement from the fiducial star is within the uncertainties in the measured offset . what are the stars in the environment of sn 1954j / v12 ? in particular , are the characteristics of star 4 consistent with an @xmath0 car - like star and the survivor of sn 1954j / v12 , as suggested by the spectrum ?
to provide possible answers to these questions , we place the four stars ( from the magnitudes in table 2 ) on a [ ( @xmath52 ) , ( @xmath53 ) ] color - color diagram , shown in figure 5 . also illustrated are the intrinsic color tracks for dwarfs and supergiants from drilling & landolt ( 2000 ) , as well as the reddening vector for the cardelli , clayton , & mathis ( 1989 ) reddening law .
we note that , for these colors , the reddening vector runs parallel to the stellar tracks . at their exposure levels
, we would expect the acs images to be sensitive generally to supergiants and the bright end of the main sequence .
stars 1 and 2 could be blue supergiants experiencing local extinction @xmath54 mag
. however , their absolute magnitudes , corrected for this extinction and the true distance modulus , would then be an implausible @xmath55 .
these two stars could possibly be reddened , luminous yellow ( g - type ) supergiants ; in this case , the required extinction would be @xmath56 mag , with @xmath57 , @xmath58 , and @xmath59 mag . however , such yellow supergiants , or so - called hypergiants , are relatively rare ( it seems even less likely that _ both _ stars would be hypergiants ) . alternatively
, both stars 1 and 2 have colors consistent with early - m supergiants ( at least for @xmath53 ; the transformation to @xmath52 appears to be too red , but the uncertainty in this color for both stars is large and still consistent with the supergiant track ) , assuming low reddening ( the galactic contribution to extinction toward ngc 2403 is only @xmath60 mag ; schlegel , finkbeiner , & davis 1998 ) , and we consider this the more likely case . star 3 could be a blue star with @xmath61 mag ; its reddening - free colors would be @xmath62 and @xmath63 mag ; its absolute magnitude , @xmath64 , would imply that it is a hot main - sequence dwarf . however , star 3 also has colors consistent with early - a to early - f type with little reddening ( although the star s locus is offset in color by @xmath65 from the stellar tracks in figure 5 ) . additionally , this star is @xmath310.7 mag fainter in f658n than in f606w , implying a relatively high h@xmath20 absorption equivalent width , consistent with an a - type star ; using the stsdas package synphot , together with synthetic spectra from the bruzual - persson - gunn - stryker ( bpgs ) spectral atlas , we find that the flight - system colors for this star are most consistent with those of an early - to - mid a star .
its inferred absolute magnitude is then only @xmath66 . with little reddening , star 4
has colors generally consistent with spectral type g ( although its locus in figure 5 is also offset in color by @xmath65 from the stellar tracks ) .
based on its observed brightness , it would have to be a supergiant .
however , we would still need to account for the presence of the broad h@xmath20 line from a g - type supergiant .
the yellow supergiants ( hypergiants ) , such as @xmath67 cassiopeiae ( e.g. , lobel et al .
2003 ) , do show h@xmath20 emission associated with a massive wind .
however , the emission profile for @xmath67 cas is quite different in shape , velocity width , and equivalent width from what we see in the keck spectrum for star 4 .
additionally , @xmath67 cas , spectroscopically of late - f or early - g type , has @xmath682 mag ( e.g. , joshi & rautela 1978 ) and @xmath57 mag ( e.g. , lobel et al . 2003 ) , whereas star 4 , as a g - type star , corrected only for galactic extinction and the distance modulus , would be too faint , at @xmath69 mag .
we therefore consider the possibility that star 4 is a yellow hypergiant to be unlikely .
star 4 , however , also could be a reddened early - type supergiant .
we have already pointed out the similarity of its spectrum to that of @xmath0 car , so it is plausible for us to assume that star 4 could be a lbv seen through appreciable extinction , as is the case for @xmath0 car . even with the color offset from the stellar tracks , the star s locus in figure 5 does , in fact ,
appear consistent with a blue star experiencing an extinction of @xmath70 mag .
we note that it is reasonable to assume that the reddening is variable across @xmath3118 pc at the sn site .
unfortunately , we do not possess another observed color ( such as @xmath71 ) or stellar spectral features for these four stars to help break some of the degeneracy between the stars possible intrinsic colors and the amount of reddening in the environment .
all four stars are somewhat redder in @xmath52 and/or bluer in @xmath53 than we might expect for the inferred stellar types , possibly due to underestimation of the uncertainties in the observed colors and to errors in the color transformation .
( for stars 1 and 2 using , for example , synphot and the bpgs atlas , we can transform the flight system colors to @xmath722.3 and @xmath73 mag , putting the stars closer to the supergiant track . )
although we can not readily account for the color offset for star 3 , at least some of the offset for star 4 ( @xmath74 mag ) from the supergiant track likely results from contamination of the continuum light in the broad f606w band by the significant emission at h@xmath20 .
we can explore this further using synphot : normalizing a 32,000 k blackbody ( or , equivalently , a late o - type or early b - type supergiant ) spectrum to an unreddened f814w magnitude of 19.8 ( absolute magnitude @xmath75 for a luminous supergiant at the distance of ngc 2403 ) , then applying a reddening @xmath76 mag ( @xmath77 mag ) , and , finally , adding a synthetic ( gaussian - shaped ) h@xmath20 emission line with the characteristics of the observed emission line in figure 2 and the observed f658n magnitude for star 4 in table 2 , results in a good fit to the observations , to within the uncertainties , for the f475w , f606w , and f814w magnitudes of 24.13 , 23.09 , and 22.14 , respectively . without the h@xmath20 emission in the model , the f475w and f814w magnitudes , of course , remain relatively unchanged ; however , the f606w magnitude is then 23.24 . following the transformations in sirianni et al .
( 2005 ) , we find @xmath78 and @xmath79 mag .
( the @xmath52 and @xmath53 colors , 0.95 and 1.30 mag , respectively , for the model derived using synphot are unaffected by the h@xmath20 emission , which is outside the @xmath18 passband . )
these colors now place the locus of star 4 much closer both to the supergiant track and to being along the vector for a reddened luminous blue star .
the h@xmath20 emission clearly leads through the bandpass tranformations to a redder @xmath52 and a bluer @xmath53 color ( exactly the trend seen in figure 5 ) than without the emission . in light of these arguments
, we consider it quite plausible that , like @xmath0 car , star 4 is a very luminous o - type or early - b star hidden behind a dusty ( @xmath70 mag ) shell or lbv - like nebula .
we argue that this is the most likely scenario , given the star s brightness , color , and the characteristics of the h@xmath20 line emission . the shell , analogous to @xmath0 car s homunculus , was possibly ejected during the 1954 super - outburst .
we therefore assume the unreddened colors for the model star , @xmath80 and @xmath81 mag , and the model extinction - free absolute magnitude , @xmath82 . including the measured uncertainties in the corresponding flight - system f475w@xmath83f606w and f606w@xmath83f814w colors and f606w magnitude from table 2 , we can place star 4 in color - magnitude diagrams shown in figure 6 , along with the possible properties of the other three stars in the environment . for comparison
we show in figure 6 stellar evolutionary tracks from lejeune & schaerer ( 2001 ) , with enhanced mass loss , for a range of stellar masses . at the position of sn 1954j / v12 in the galaxy ( @xmath84 from the nucleus , or galactocentric distance @xmath85 kpc ) the metallicity could be somewhat subsolar : from a study of the metallicity gradient in ngc 2403 by garnett et al .
( 1997 ) , the value of log(o / h)+12 at this distance is 8.6 dex , which is a factor 1.58 below solar metallicity ( the solar log(o / h)+12 abundance is 8.8 dex ; grevesse & sauval 1998 ) . however , the first available subsolar model tracks from lejeune & schaerer are for @xmath86 , which is a factor 2.5 below solar metallicity ( @xmath87 ) , and we consider this too low to adequately represent the sn 1954j / v12 environment .
therefore , we choose the model tracks with solar metallicity ( although we note below the ramifications if we used the @xmath86 tracks instead ) .
we conclude from figure 6 that stars 1 and 2 are either yellow supergiants of @xmath88 or , more likely , lower - mass ( @xmath89 , or @xmath90 for the @xmath86 track ) red supergiants , whereas star 3 could be a blue main - sequence dwarf with a range of possible ( likely , low ) masses .
more importantly , we conclude that star 4 has a mass @xmath91 ( or @xmath92 for the @xmath86 track ; the mass is most constrained in figure 6b ) .
unfortunately , the photometric uncertainties do not allow us to more rigorously constrain the mass of star 4 . however , the spectral and photometric similarities to @xmath0 car , and its implied high absolute @xmath18 magnitude , argue that the star s mass is likely closer to the upper range of massive stellar tracks shown in figure 6 .
we therefore find it highly suggestive that this star is the very luminous , massive survivor of the sn 1954j / v12 eruption .
what about the stellar environment of sn 1954j / v12 , relative to the environments of @xmath0 car and other lbvs ?
how likely is it that these four stars constitute an association of massive stars spread over @xmath93 ( @xmath94 pc ) ?
the star @xmath0 car is a member of the populous ob association trumpler 16 , part of a much larger ( @xmath95 pc ) site of intense massive star formation in the carina nebula ( walborn 1995 ) .
p cygni appears to be a member of a rich young cluster @xmath96 pc northeast of the cluster ic 4996 and part of the larger cyg ob1 association ( turner et al .
sn 1961v also sits in an environment of luminous , presumably massive supergiants over @xmath97 pc ( filippenko et al .
1995 ; van dyk et al .
although not an @xmath0 car analog , sn 1987a arose from a high - mass blue supergiant and occurred in a loose , young ( age @xmath3112 myr ) cluster ( panagia et al .
2000 ) ; its closest known companion is a coeval b2iii star ( scuderi et al .
however , the lbv ag car does not seem to be associated with any cluster of luminous , massive stars over @xmath98 pc ( hoekzema , lamers , & van genderen 1992 ) .
therefore , although the tendency is for the lbvs and @xmath0 car analogs to occur in rich , massive clusters spread over many pc , a range of environments exists for very massive stars .
it is possible , then , for the four stars in the sn 1954j / v12 environment to be a sparse cluster of roughly coeval massive stars ( see figure 6 ) .
what about the dusty nebula ?
the profile of the source in the f658n image is unresolved ( @xmath311.5 wfc1 pixels fwhm ) , which implies a radius @xmath99 pc for the nebula . we can further constrain the size of the nebula from the luminosity in the h@xmath20 emission line and recombination theory , following wagner et al .
( 2004 ) . assuming an average @xmath100 erg s@xmath7 @xmath43 @xmath7 and equivalent width @xmath31240 for the line , and correcting for extinction using the relation @xmath101 ( viallefond & goss 1986 )
, we derive an estimate of the h@xmath20 luminosity for the nebula @xmath102 erg s@xmath7 at the distance of ngc 2403 . assuming the same properties for this nebula as did wagner et al .
( 2004 ) for the nebula around sn 2000ch , we find @xmath103 pc , comparable to ( within a factor of two of ) the size of @xmath0 car s homunculus ( humphreys & davidson 1994 ) .
through imaging and spectroscopy obtained with the keck 10-m telescopes and imaging with the acs on - board _ hst _ , we have confirmed that the precursor star of sn 1954j / v12 in ngc 2403 has survived a super - outburst , analogous to the 1843 eruption of @xmath0 car .
it is not a sn , in the classical sense , but an `` impostor '' masquerading as a sn .
furthermore , the acs images , in combination with the keck spectroscopy , have revealed this quite rare @xmath0 car analog as a highly luminous , massive star ( @xmath2 ) , surrounded by a dusty , homunculus - like nebula .
( based on this conclusion for sn 1954j , we also now consider it even more likely that the @xmath0 car - like star detected in the sn 1961v environment by chu et al .
2004 is , in fact , the survivor of that event . )
the fact that the sn 1954j / v12 super - outburst survivor has changed little in brightness and color over the last eight years implies that the star is now in a relatively quiescent state .
its behavior in the recent past suggests that monitoring of the object should continue into the future , in order to understand more about this important , but rare , phenomenon in the evolution of massive stars .
we are grateful to soeren larsen with assistance in the calibration of his images .
doug leonard and adam riess provided assistance during the keck imaging run .
we thank roberta humphreys for illuminating discussions .
this publication makes use of data products from the two micron all sky survey , which is a joint project of the university of massachusetts and the ipac / california institute of technology , funded by nasa and nsf .
we have also utilized the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with nasa .
the work of a.v.f.s group at uc berkeley is supported by nsf grant ast-0307894 , as well as by nasa grants ar-9953 and ar-10297 from the space telescope science institute , which is operated by aura , inc .
, under nasa contract nas5 - 26555 .
some of the data presented herein were obtained at the w. m. keck observatory , which is operated as a scientific partnership among the california institute of technology , the university of california , and nasa ; the observatory was made possible by the generous financial support of the w. m. keck foundation .
is grateful for a miller research professorship at u.c .
berkeley , during which part of this work was completed .
lcccccc 1996 oct .
20 & 2450377.1 & & & 21.31(30 ) & 21.06(09 ) & 20.04(09 ) + 1997 oct .
13 & 2450734.7 & 21.67(06 ) & 22.21(05 ) & 21.74(05 ) & 20.98(05 ) & 20.14(07 ) + 1999 feb .
1421 & 2451227.0 & 22.5(3 ) & 22.7(2 ) & 21.9(3 ) & 21.1(2 ) & 20.9(2 ) + 2004 aug .
17 & 2453235.0 & & 22.60(23 ) & 21.67(15 ) & & 20.12(12 ) + cccccccc 1 & 23.99(33 ) & 22.25(12 ) & 20.38(08 ) & 22.10(15 ) & 24.74(33 ) & 22.72(12 ) & 20.40(08 ) + 2 & 24.08(34 ) & 22.30(12 ) & 20.56(07 ) & 22.20(17 ) & 24.85(34 ) & 22.76(14 ) & 20.58(08 ) + 3 & 23.01(14 ) & 22.81(17 ) & 22.82(44 ) & 23.49(42 ) & 23.07(14 ) & 22.83(17 ) & 22.81(44 ) + 4 & 23.89(35 ) & 22.88(19 ) & 22.22(26 ) & 21.00(05 ) & 24.29(35 ) & 23.08(21 ) & 22.21(26 ) +
|
we have confirmed that the precursor star of the unusual supernova 1954j ( also known as variable 12 ) in ngc 2403 survived what appears to have been a super - outburst , similar to the 1843 great eruption of @xmath0 carinae in the galaxy .
the apparent survivor has changed little in brightness and color over the last eight years , and a keck spectrum reveals characteristics broadly similar to those of @xmath0 car .
this is further suggested by our identification of the actual outburst - surviving star in high - resolution images obtained with the advanced camera for surveys on the _ hubble space telescope_. we reveal this `` supernova impostor '' as a highly luminous ( @xmath1 mag ) , very massive ( @xmath2 ) eruptive star , now surrounded by a dusty ( @xmath3 mag ) nebula , similar to @xmath0 car s famous homunculus .
|
cerebellum deserves more extensive studies than was conventionally realized .
historically , cerebellum was thought of as mainly a motor control organ , while recent researches reveal that it has many other functions , and it has more intimate connections to other parts of the brain @xcite .
histamine ( ha ) , a neurotransmitter or neuromodulator in the brain , plays an important role in the functions and interactions of various parts of the brain , and also in the studies of these functions and interactions .
for instance , the neuroanatamic researches revealed the existence of the hypothalamus - cerebellum histaminergic path @xcite , and shows that hypothalamus has great influence on the cerebellar activities and hence plays an important role in coordinating the functions of the body and viscera .
besides , ha has the possible role of controlling the cerebellar circulation , and ha receptors are also found in the neurons of the cerebellar cortex @xcite . cerebellum is made up of the outer dark matter ( cortex ) , inner white matter , and three pairs of deep nuclei lying in the heart of the white matter .
these nuclei are the fastigial nucleus fn , interposed nucleus in , and dentate nucleus dn .
the afferent fibres to the cerebellum are mainly from the vestibule , spinal cord , and the cerebral cortex .
they form synaptic connections with the neurons in the cerebellar deep nuclei and cortex .
the synapses of purkinje cell ( pc ) make the efferent fibres of the cerebellar cortex , they are mainly projected into the deep nuclei , then the neurons there stick out fibres which make the cerebellar output ; a small amount of pc synapses are directly projected into the vestibular nucleus .
the cerebellar cortex can the divided into three layers : the ( out - most ) molecular layer , pc layer , and the granular layer .
it contains three kinds of afferent fibres ( musciform fibre mf , crawl fibre cf , and monoaminergic fibre ) , and five kinds of neurons ( pc , granular cell gr , basket cell ba , star - like cell st , and golgi cell go ) .
hence we see that the cerebellar afferent fibres and the intermediate neurons , with pc acting as the core , constitute the basic neural circular that is responsible for the sensory function of the cerebellar cortex , and the cerebellar cortex together with the deep nuclei undertake various functions of the cerebellum
. in our laboratory there has been studies of ha s effects on certain neurons of the cerebellum , such as the granular cells @xcite , while ha s effect on the neurons in the cerebellar cortex is not yet intensively studied .
considering that pc is the only efferent neuron of the cerebellar cortex , we are going to investigate ha s effect on the electric activities of pc , based on our accumulated experiences in studying the influences of aminergic materials such as norepinephrine ( na ) and serotonin ( 5-ht ) on the spontaneous and induced discharge activities of the cerebellar pc ; so as to learn more about the role of aminergic afferent system in the process of information treatment in the cerebellar cortex .
we use for our experiments 19 mature sd rats ( 200 - 250 g ) . anaesthetize a rat by injecting betchloramines hydrochloride ( 4mg/100 g ) into the abdominal cavity ,
take out the cerebellum right after cutting the head , wash the cerebellum with frozen artificial cerebrospinal fluid ( acsf , @xmath1c ) , stick it onto the operating table of a microtome ( at which the cerebellum is soaked in frozen acsf ) , and cut a 400@xmath2 thick of arrow - like slice from the vermis .
the process of making the slice should be done within 20 minutes .
then move the so obtained slice into a recording trough , and begin the experiment after 15 minutes of hatching .
the recording trough is continuously irrigated ( 3ml / min ) by acsf ( 33@xmath3 ) , and is aerated with the mixed air of 95%o@xmath4 + 5%co@xmath4 .
the concentrations of various elements in acsf are ( mmol / l ) : nacl 124 , kcl 5 , kh@xmath4po@xmath5 1.2 , mgso@xmath5 1.3 , cacl@xmath4 2.4 , nahco@xmath6 26 , glucose 10 . put a tiny glass electrode ( filled in with colored conducting fluid ) at the pc layer of the x leaflet of the cerebellar cortex to make out - cell records of pc s discharge activity .
this is because the x leaflet received the least mechanical wound in making the slice , and hence the cerebellum slice should be so put in the recording trough that the x leaflet be well hatched by the acsf and the mixed air .
we base on the following criteria to single out pc discharge signals : ( 1 ) _ position of the electrode _
: pcs in the cerebellar cortex are of linear type , namely , the cell bodies concentrate at one end and the dendrites stick to the other end , thereby form a pc layer in the cerebellar cortex .
we put the recording electrode at the out side of the pc layer and near to the molecular layer , therefore there is very little chance to catch a discharge signal of the granular cell ; ( 2 ) _ discharge wave shape _ : the pc action potential has large magnitude ( [email protected]@xmath8 ) and thick contour , moreover , the effective recording distance of pc discharge is 50@xmath9 m , much longer than that of the granular cell ( 20 - 30@xmath10 m ) @xcite ; ( 3 ) _ cell numbers _ : in the cerebellar cortex there are much less go , st , and ba than pc , and hence the discharge signals from these cells may be ignored . because the nerve cells in the brain have the character of continuously producing impulses , and often at the frequency of tens of times per second , the physiological status of a neuron is usually symbolized by its discharge frequency . under the environmental stimulation
, the discharge frequencies of central neurons will variate , and the intrinsic potentials will also deviate .
hence we magnify the pc discharge signal , choose a relatively small time constant for the recording system , so as to differentiate the input signal and transform the slowly increasing wave into a sharp high peak wave , then use a pulse discriminator to convert the peak signal into ttl pulse and input it into the computer , and then draw the post - stimulus histogram . thus we can describe pc s reaction to chemicals by the post - stimulus histogram of its discharge frequency .
all the drugs here are contemporarily compounded using acsf , and are used to irrigate the cerebellum slice separately . when doing the experiments of studying the receptor mechanism of ha s effect on pc , the slice is continuously irrigated by the receptor antagonist for over 10 minutes before by ha .
because intermittent discharge is often due to the ill state of pc or indistinguishability of cell discharge signal from other signals , we choose in 29 cerebellum slices 59 cells with relatively stable spontaneous discharge for our study .
their spontaneous discharge frequencies range from 4.48 to 78.32 hz .
we find that of the 59 cells , 83.1% ( 49/59 ) are affected by ha , the other 16.9% ( 10/59 ) show no response . and of the 49 responsive ones , 87.8% ( 43/49 ) are excited , the other 12.2% ( 6/49 ) are inhibited .
( see fig .
1 ) therefore ha s effect on pc is mainly excitative . from figs .
1b , 1c , we can see that there is an evident increase of the variation of the cell s discharge frequency as the irrigating concentration of ha increases .
l b l c another important observation is that whether or not a pc shows response to ha may have some connection with its spontaneous discharge frequency .
( cf . fig .
2 ) the spontaneous discharge frequencies of the 59 tested cells are 4.48 - 78.32 hz , while that of the 49 cells which show response to ha is [email protected] hz ( m@xmath7sd ) , this is significantly lower than the average frequency ( [email protected] hz ) of the 10 cells unaffected by ha . however , whether a cell is excited or inhibited by ha seems to have nothing to do with its intrinsic frequency .
the average frequencies of the 43 excited cells and of the 6 inhibited cells are [email protected] and [email protected] hz , respectively , which are not much different ( [email protected] , t test ) .
this differs from how ha s effect on the cerebellar granular cells is related to their intrinsic frequencies @xcite
. we also find that among the 49 responsive cells , those who have higher intrinsic discharge frequencies are more sensitive to ha ( there are 8 responsive cells with intrinsic frequencies higher than 30 hz , and 7 of them show response to ha at the lowest ha concentration of less than 30@xmath10mol / l ) .
l a l b it is known that there are three sub - types of ha receptors in brain : h1 , h2 , and h3 , therefore it is necessary to study the receptor conducting mechanism of ha s effect on pc . since ha s main effect on pc
is excitative , we make study for the excitative effect . for this purpose
we observe the influences of triprolidine ( hyper - specific antagonist for h1 receptor ) and ranitidine ( hyper - specific antagonist for h2 receptor ) on pc s excitative reaction to ha , and find that low - concentration triprolidine ( 0.5 - 1.0@xmath10mol / l ) could weaken ha s excitative effect on pc , while low - concentration ranitidine ( 0.9 - 1.0@xmath10mol /
l ) could weaken or even block ha s excitative effect on pc .
( cf . fig .
h3 receptor is a presynaptic self - receptor , it adjusts the presynaptic release of ha @xcite . in this research
we have not investigated its role in ha s excitative effect on pc .
l a l b l c
in this experiment we used the data - taking system : oscillometer - pulse discriminator - computer , and drawn the post - stimulus histogram , then evaluate pc s reaction to chemicals by the post - stimulus histogram of the discharge frequency .
the sampling bin width is 3 , point=900 , interval=1000 , one data represents the discharge times of a cell in one second .
since a cell discharges at very high frequency and its electric signals transmit at very high speed , it gives very rich information in one second .
we know that the information carried by a neuron is represented by its frequency distribution over the action potential , therefore if we can obtain the frequency spectrum , and make the combined analyses of the frequency and power spectra , we will get greater amount information .
besides , what we take in this experiment is out - cell record of a cell s discharge activities , and pc is in connection with various other cells in the cerebellum slice , the measurement in such case may be disturbed by the environment , and the excitative or inhibitive reaction of pc may actually be a combined result of various influences including those from the environment .
these are to be improved in future studies .
histamine is an important neurotransmitter or neuromodulator , and acts as an inter - cell messenger .
ha can hardly penetrate the blood - brain barrier , hence must be produced by the histaminergic neurons .
it can be released through depolarization and by means of calcium dependence .
the hypothalamus - cerebellum histaminergic fibres in the brain are casted by the hypothalamus tuberous mastoid nucleus ( ha pericaryon ) into the cerebellar cortex and the cerebellar deep nuclei ( ha nerve end ) @xcite .
cerebellum - hypothalamus histaminergic system can bring about many functions via the inter - cell messenger ha .
the hypothalamus - cerebellum histaminergic projecting fibres end at the cerebellar cortex as spreading multi - layer fibres , and their nerve ends mainly spreadingly adjust the functions of the peripheral neurons in the varix form ; these are much like the serotoninergic and norepinephrinergic afferent fibres of the cerebellum @xcite .
purkinje cell is the only kind of efferent neuron of the cerebellar cortex , it is a deformation of the multi - pole cell , and has very strong dentrite .
our research shows that pc s discharge activity is significantly affected by ha . from our observations in sec .
iiia , and associating how the reaction of cerebellar granular cells to ha is related to their intrinsic discharge frequencies @xcite , we speculate that the cerebellar afferent fibres might be mainly responsible for adjusting the basic discharge levels of the cerebellar neurons : as the discharge frequency of a target cell lies in a certain region it loses sensitivity to ha ; and as the discharge frequency lies out of this region the cell reacts to ha , and thereby its frequency tends to that region .
moreover , the high - frequency region is a low - sensitivity region to ha , while for the responsive cells , those with higher frequency are highly sensitive to ha .
we therefore speculate that there exists a critical frequency , a cell with intrinsic frequency near to this critical point will significantly change its states by only a small amount of ha , while as its frequency goes well above this point , it becomes stable . in connection with the influence of na and 5-ha on the spontaneous and induced electric activities of cerebellar pc @xcite
, we go further to speculate that histaminergic and other aminergic afferent systems might , through their coordinations and/or antagonism , adjust the synaptic transmission efficiency of mf - pf ( parallel fibre)-pc and cf - pc , or adjust pc s sensitivity to signals from mf and cf , and thereby take part in the the global process of the sense movement of the cerebellar neuron net . that many kinds of neural active materials mutually act on the neuron is a common pattern of signal transmission of the nerve system ; such mutual action may happen at the presynapse , postsynapse , or even postreceptor level .
ha s effect on the cerebellar pc is to adjust the excitativity level of the neuron and the neuron s sensitivity to input information from outside of the cerebellum , and adjust the sleep or awake state of the cortex .
the aminergic afferent fibres have synaptic or non - synaptic chemical transmission effect on the cerebellar cortex , their non - synaptic transmission may not encode the fast phase - like information , but hyperfinely adjusts the membrane potential and basic discharge level or the target neuron . of the three sub - types of ha receptors in the brain ( h1 , h2 , and h3 ) ,
h3 receptor is a presynaptic self - receptor .
the acting mechanism of h1 receptor is through some sub - type of g protein to activate plc , hydrolize 4,5-pip@xmath4 to be 1,4,5-ip3 and dg , these two secondary messengers go further to trigger various biological effects , such as stimulating endoplasmic reticulum to release ca@xmath13and thence trigger the ion channels @xcite .
the h2 receptor brings effect mainly via g protein - ac - camp : ha combines with the h2 receptor of the cell membrane , the h2 receptor is then deformed and triggers g protein to give off its @xmath14 and @xmath0 sub - radicals , and on the @xmath15 sub - radical gdp is replaced by gtp , then ac is activated and catalyzes atp to transform to camp , camp acts as a secondary messenger and triggers a series of reactions , such as activating pka and then triggering ion channels , or directly triggering ion channels . because many highly efficient enzymes take part in these reactions , small signals are magnified step by step .
it is usually thought that h1 and h2 receptors conduct ha s excitative and inhibitive effects on neurons respectively @xcite .
however , this is in contradiction with our report in sec .
iiib that low - concentration h1 receptor antagonist could weaken ha s excitative effect on pc , while low - concentration h2 receptor antagonist could weaken or even block ha s excitative effect on pc .
this makes us to conjecture that _ both _ h1 and h2 receptors involve in ha s excitative effect on pc , and h2 receptor is the _
main _ conductor .
there has been report that ha s excitative effect on the granular cells of the cerebellar cortex and on the cells of the vestibular inner - side nucleus is mutually conducted by h1 and h2 receptors .
histochemistry researches also reveal that in the cerebellar cortex of rat the density of h2 receptor is higher than that of h1 receptor .
all these partly support our conjecture .
this may suggest that the h2-related signal transmission chain in cell differs from one type of neuron to another , and such difference determines whether the h2 receptor conducts excitative or inhibitive effect ; but the particular mechanism of signal transmission is still to be studied . to systematically study the biophysical or biochemical mechanisms of the reactions of cells , it is often helpful to make mathematical simulations of the reactions . here
we give a simulation for the relation between pc s ( excitative ) reaction to ha ( measured by its maximum variation of discharge frequency under the action of ha at a certain irrigating concentration ) and its intrinsic discharge frequency @xmath16 . to be concrete , we choose a fixed frequency variation @xmath17 , and denote by @xmath18 the irrigating concentration of ha at which the maximum change of a cell s frequency would be @xmath17 , then study the relation between @xmath18 and @xmath16 . for this purpose
we first look at how the maximum frequency @xmath19 of a cell under the action of ha is related to ha s irrigating concentration @xmath20 . as we noted in sec .
iiia , pc s reaction to ha has something to do with its intrinsic discharge frequency : high - frequency cells are insensitive to ha , while for the cells affected by ha , the higher - frequency ones are highly sensitive to ha , and at low intrinsic frequencies @xmath19 increases steadily with @xmath20 .
we therefore assume that @xmath19 is an @xmath21-like function of @xmath20 : @xmath22 as @xmath23 , @xmath24 is just the intrinsic frequency @xmath16 : @xmath25 according to our above explanations , we write @xmath26 from eqs .
( [ f0],[fu0 ] ) we can derive the @xmath18-@xmath16 relation : @xmath27 here the physiological meaning of @xmath28 is the maximum discharge frequency of the cell , and @xmath17 is an arbitrarily chosen minimum frequency variation that can be observed in the experiment .
it is straightforward to check that @xmath29 has a minimum value at an intermediate @xmath16 , just as the experimental results show .
( [ u0 ] ) is much like the relation in the hopfield model . but
further study of the connection to the hopfield model would require more detailed knowledge about the relation between @xmath19 and @xmath20 , which is not possible in the present experiment , for pc can not live through a long enough time to allow tests for various ha concentrations ; hence the simulation here is merely a rough one .
( [ f],[u0 ] ) show a pattern very similar the characteristic frequency of the cerebrum ( the @xmath0 region of around 40 hz ) : there exist a particular frequency at which the target cell is highly sensitive to ha , and well above this frequency the cell loses sensitivity to ha . for pc
this insensitive region is 25.00 - 46.24 hz , with the average [email protected] , which is close to the @xmath0 region .
these suggest that cerebellum might also exhibit some sort of characteristic frequency to external stimulations .
further studies of such characteristic frequency would necessarily take into account of the interactions of various cell , such as in the hr model @xcite ; these we hope to accomplish in the future .
the author owes special thanks to prof .
jiang - jun wang and prof .
bing zhu for valuable instructions , to le tian and jie ma for collaborations in this experiment , and to feng dong , qin xi , and guo - ning hu for kind help and suggestions .
|
the effect of histamine ( ha ) on the electric activities of purkinje cell ( pc ) is studied on the cerebellum slice .
we find that : ( 1 ) ha s main effect on pc is excitative ( 72.9% ) ; there are also a small amount of pc showing inhibitive ( 10.2% ) or no ( 16.9% ) response to ha .
( 2 ) different from the conventional opinion , ha s excitative effect on pc is mutually conducted by h1 and h2 receptors ; the antagonist for h1 receptor could weaken ha s excitative effect on pc , while the antagonist for h2 receptor could weaken or even block the excitative effect of ha on pc .
( 3 ) pc s reaction to ha is related to its intrinsic discharge frequency ; there exists a frequency at which pc is highly sensitive to ha , and well above this frequency pc becomes stable against ha .
these results indicate that the histaminergic afferent fibre can adjust pc s electric activities by releasing ha , and thereby influence the global function of the cerebellar cortex ; and that just like the @xmath0 region of cerebrum , cerebellum may also have some sort of characteristic frequency .
key words : cerebellar cortex ; purkinje cell ; histamine ; receptor ; cerebellum slice 2
|
the key to understanding the star formation history of galaxies is determining the age of their underlying stellar populations and the amount chemical evolution that those populations have undergone .
historically , these issues have been addressed separately and by morphological type .
for example , early - type galaxies are considered have the simplest modes of star formation , the oldest ages ( i.e. primordial ) and relatively passive evolution as indicated by their similarity in color , morphology and kinematic properties ( bernardi 2003 , tremonti 2004 , cool 2006 , smolcic 2006 ) .
very little current or recent star formation ( bernardi 2001 ) , plus low amounts of molecular gas ( huchtmeier , sage & henkel 1995 ) , reinforces the hypothesis that a majority of their stars have their origin at high redshift ( larson 1975 , kodama 1998 , jimenez 1999 , van dokkum & stanford 2003 ) .
work on the star formation history of galaxies is dominated by studies of rich clusters because 1 ) they are located in the same region of position and velocity space easing observational constraints , 2 ) they are easily detected in surveys at large redshift and are useful signposts for the study of evolutionary effects and 3 ) they are impacted by similar , if not complicated , environmental effects . since
rich clusters are dominated by elliptical and s0 galaxies , then they are ideal laboratories for the understanding of the simplest systems , but the discovery of the butcher - oemler effect ( butcher & oemler 1984 , poggianti 2004 ) means that rich clusters are also interesting arenas to study starburst phenomenon .
tracers of age and metallicity in even simple systems , such as cluster ellipticals , are complicated by the fact that , unlike globular clusters , the stellar populations in ellipticals , are a mixture of stars of many ages at various stages of chemical evolution ( see studies that resolve the stellar populations in ellipticals by hst , gregg 2004 ) .
this means that any information obtained on galaxies , outside the resolution range of space imaging , will be luminosity weighted and observational constraints are such that these data must be interpreted using the guidance of star formation models . while model guided results always run the gamut of dependences on various assumptions , the range of galaxy types and lookback time has successful allowed a critical examination of these assumptions ( renzini 2006 ) .
in addition , the star formation and chemical evolution histories of galaxies can be examined through the use of many parameters based on their integrated light ( e.g. broadband colors versus line spectroscopy ) , each providing their own unique picture of the underlying stellar populations .
the determination of a galaxy s age and metallicity ( meaning not only the mean quantities , but the distribution of age s and metallicities within a galaxy s stellar population ) is an observationally intense problem and suffered in early studies from the problem that broadband colors are coupled with respect to age and metallicity ( the so - called age - metallicity degeneracy problem , see worthey 1999 ) .
however , in the last decade , three approaches have successful broken the age - metallicity degeneracy .
the first is the use of colors across a wide wavelength region ( e.g. far - uv to near - ir , see kaviraj 2006 ) which bridges regions of the spectrum sensitive to many different regions of the hr diagram .
the second method is to use various spectral features sensitive to age and metallicity in a semi - independent manner ( gonzalez 1993 ) .
semi - independent in a fashion that meaningful values can only be extracted by comparison to stellar population models .
there is nothing new to the use of models to determine stellar population parameters , for example , globular cluster ages can only be extracted by comparison to stellar evolution tracks .
but an additional complications arises in that the stellar populations in galaxies are neither homogeneous in age nor metallicity ( if our own galaxy is a typical example ) .
the third technique is the one pioneered by our team using narrow band filters combined with a robust principal component analysis ( pca ) of multi - color space derived from spectroevolutionary ( sed ) models ( rakos & schombert 2005a ) .
this technique makes use of the effects of age and metallicity on the continuum emission resulting from the integrated light of the underlying stellar population . for the past decade
, the technique of choice to determine age and metallicity in galaxies has been judicious choice of certain spectral line indices ( i.e. the lick sequence , worthey & ottaviani 1997 , trager 2000 ) .
these studies sample particular lines sensitive to metallicity ( e.g. , fe and mg ) combined with other couplets to find age ( e.g. the balmer lines ) .
the grid of age and [ fe / h ] is model dependent , but the two values are distinct and have a solvable degeneracy problem . while line measurements would appear , at first glance , to be a superior technique since their are actually measuring the parameters to investigate the chemical evolution of stellar populations ( e.g. fe , mg ) , line strengths do suffer a number of disadvantages . increased information on various lines paints a confusing picture .
for example , fe strengths show very little dependence on galaxy mass ( fisher , franx & illingworth 1997 ) , in contrast to expectations from the color - magnitude relation .
there are sharp dependences between @xmath2 elements , such as ca and mg , and the main source of free electrons in stellar photospheres , fe .
for example , sanchez - blazquez ( 2006b ) find no correlation between age and metallicity when the mg@xmath3 index is used , but clear correlations are found with ca or fe indices .
this leads to difficulties in forming an overall picture of galaxy formation as any model of stellar populations does not necessarily desire the value of individual elements as input parameters , but rather their combined effect on stellar atmospheres as this reflects into their position on the hr diagram .
recently , the values for galaxy age extracted from the lick technique has come in conflict with age determination by other methods .
numerous studies of cluster ellipticals using the lick system that find many examples of ellipticals with relatively young mean ages ( less than 6 gyrs , sanchez - blazquez 2006 ) .
in contrast , our previous work using continuum colors , flux values determined from regions of the near - blue portion of a galaxy s spectra , find very few ellipticals with ages less than 10 gyrs . and
recently , ellipticals have also been determined to have old ages in a study using far - ir indicators of age ( bregman , temi & bregman 2006 ) .
some of age dispersion may be due to contributions from small amounts of young stars , significant in their effect on colors , but minor in their fraction compared to the total stellar population ( i.e. , the ` frosting ' effect , trager 2000 ) .
however , again , this will be an advantage for integrated continuum colors which more fully take into account the whole stellar population in a galaxy and will not suffer from aperture effects that may distort spectral estimates .
the dispersion in ages for cluster galaxies is a major inhibitor in determining the correct scenario for galaxy formation and evolution .
while a range of ages can be supported by various methods of galaxy formation , purely old galaxy ages imply high redshifts for galaxy formation and relatively simple evolutionary scenarios . to further explore the issue of galaxy age
, we developed a technique parallel to line indices work that focuses on the continuum region of the spectra around 4000 .
we have refined our calculations of age and metallicity from our narrow band colors and incorporated a robust technique to determine those values from the data ( see rakos & schombert 2005a ) .
this paper will apply these new methods to examine the history of galaxies in a nearby rich cluster , a1185 , as compared to our typical rich cluster for extragalactic studies , coma .
our goal is to further refine the spectrophotometric classification of galaxies on our system and determine the age and metallicity properties of cluster galaxies with respect to their stellar mass and cluster environment .
the rich cluster a1185 is a bautz - morgan type ii cluster with a richness class of 1 .
it is located at a redshift of 0.033 and is approximately 80 arcmins in diameter .
a1185 is a dynamically young cluster based on its projected density and velocity distribution .
it composed of several distinct ( spatially as well as kinematically ) subgroups and a core which has several subclumps ( mahdavi 1996 ) .
our photometry includes only those galaxies on the region that is one mpc in radius centered at the x - ray luminosity peak .
this region also contains the top 10 brightest cluster members and will , presumable , become the cluster core when virialization is complete . a1185 was observed on 08 jan 2005 , using the 2.3 m bok telescope of the steward observatory located at kitt peak arizona .
the image device was 90prime , a prime focus wide - field imager using a mosaic of four 4k by 4k ccds resulting in a field of view of 1.16 by 1.16 degrees and a plate scale of 0.45" / pixel ( williams 2004 ) .
the filter system , a modified strmgren ( @xmath4 ) system , was the same as used in our distant cluster studies ( rakos & schombert 1995 ) .
five exposures of 600 secs were made in each filter centered on ngc 3552 .
calibration was obtained through a number of spectrophotometric standards measured on each night . the modified strmgren system ( herein called @xmath5 )
is altered in the sense that the filters are slightly narrower ( by 20 ) and the @xmath6 filter is shifted 30 to the red in its central wavelength as compared to the original system .
the system we use herein is called the @xmath5 system to differentiate it from the original @xmath4 system since our filters are specific to the rest frame of the cluster that is being studied .
the @xmath5 system covers three regions in the near - uv and blue portion of the spectrum that make it a powerful tool for the investigation of stellar populations in ssp s ( simple stellar population ) , such as star clusters , or composite systems , such as galaxies .
the first region is longward of 4600 , where in the influence of absorption lines is small .
this is characteristic of the @xmath7 and @xmath8 filters ( @xmath9 = 4675 and 5500 ) , which produce a temperature color index , @xmath10 .
the second region is a band shortward of 4600 , but above the balmer discontinuity .
this region is strongly influenced by metal absorption lines ( i.e. fe , cn ) particularly for spectral classes f to m , which dominate the contribution of light in old stellar populations .
this region is exploited by the @xmath11 filter ( @xmath12 ) .
the third region is a band shortward of the balmer discontinuity or below the effective limit of crowding of the balmer absorption lines .
this region is explored by the @xmath13 filter ( @xmath14 ) .
all the filters are sufficiently narrow ( fwhm = 200 ) to sample regions of the spectrum unique to the various physical processes of star formation and metallicity ( see rakos 2001 for a fuller description of the color system and its behavior for varying populations ) .
the reduction procedures have been published in rakos , maindl & schombert ( 1996 ) and references therein .
the photometric system is based on the theoretical transmission curves of filters ( which can be obtained from the authors ) and the spectra of spectrophotometric standard stars published in the literature ( massey & gronwall , 1990 ) .
the convolution of the transmission curves and the spectra of the standard stars produces theoretical flux values for color indices of the standard stars corrected for all light losses in the equipment and the specific sensitivity of the ccd camera .
magnitudes were measured on the co - added images using standard iraf procedures and are based for brighter objects on metric apertures set at 32 kpc for cosmological parameters of @xmath15 km s@xmath16 mpc@xmath16 and the benchmark cosmology ( @xmath17 , @xmath18 ) . for fainter objects , the apertures were adapted to deliver the best possible signal to noise ratio , but always using the same aperture for all four filters .
color indices are formed from the magnitudes : @xmath19 , @xmath10 , @xmath20 , and @xmath21 [ @xmath22 .
there is a small , but measurable galactic extinction in the region around a1185 ( @xmath23 mag ) , which corresponds to @xmath24 , @xmath25 and @xmath26 .
while these corrections were made to all the datasets , their effect on the age and metallicity results ( below ) were negligible and are dominated by other sources of error .
typical errors were 0.03 mag in @xmath20 at the bright end of the sample and 0.08 mag at the faint end . after applying our photometric membership criteria to remove foreground and background galaxies
( see rakos 1991 for a full description of the procedure ) there remained 192 galaxies ranging from the brightest at @xmath27 ( ngc 3550 ) to a completeness limit of 20 mag .
of those 192 , 53 had known redshifts and all 53 galaxies were cluster members ( meaning their redshifts were within 1000 km / sec of the cluster mean redshift ) .
this confirms the high reliability of our photometric membership technique , as discussed in our previous papers .
six galaxies in our sample display radio emission and one galaxy is a liner .
our photometry system uses filters that avoid strong emission lines of star formation or liner activity , plus our colors are based on integrated colors where the stellar population dominates over any nuclear emission .
this is demonstrated by our colors for the above mentioned galaxies which are normal elliptical types under our photometric classification methods .
the three colors ( @xmath28 , @xmath20 and @xmath10 ) and the @xmath21 index ( @xmath29 ) for all 192 galaxies in a1185 are displayed in figure 1 .
this diagram is the main diagnostic tool for interpreting the narrow band colors .
it has the advantage of demonstrating that the continuum colors of galaxies is a relatively simple , to first order , distribution over a range of hubble types .
the distribution of colors in figure 1 is extremely well correlated , although the typical error is less than the width of the correlation .
changes in metallicity are a linear relation in color space ( for a constant age , see solid line for a 13 gyr population ) , thus , the scatter about a mean relationship suggests the effects of age for the red population ( see below ) and recent star formation for the blue population . as noted by smolcic ( 2006 ) ,
the narrow band colors of galaxies are nearly a one parameter family and that the locus width in principal component space is well explained by simple star formation models .
these colors are also strongly correlated with spectral indices , even emission lines , which track the current and recent star formation histories .
index ( @xmath30 ) are shown for galaxies determined to be cluster members by photometric criteria . the red population ( galaxies with @xmath31 ) is denoted by that color as is the blue population .
the solid line in each diagram displays the 13 gyr models for metallicities ranging from @xmath320.7 to + 0.4 ( schulz 2002 ) . the dotted line in the @xmath20 and @xmath21 diagrams represents the 99,000 galaxies from the sdss sample smolcic et al .
the dashed line in the @xmath28 diagram is the range in age for a solar metallicity model .
the dotted line in the same diagram displays a ` frosting ' model , i.e. adding increasing fractions of a 0.1 gyr population to a 13 gyr population .
the reddening vector is shown in the @xmath20 diagram .
the mean error bar for the sample is shown in the upper left of each panel.,width=642 ] the uniform locus in colors is particularly important for our study as the sample is selected from galaxies in a rich cluster .
the combination of dense environment and a limited range in morphological types produces an extremely red population of galaxies with a limited scatter .
figure 1 displays the obvious result that a majority of the galaxies in a1185 have typical elliptical colors ( @xmath33 and @xmath34 ) and follow an old ( greater than 10 gyrs ) track of metallicity from spectroevolutionary models ( a 13 gyr track is shown as a solid line in figure 1 for comparison ) .
we can also compare these colors with synthetic colors generated from sdss spectra ( smolcic 2006 ) .
their mean color locus are shown in figure 1 as a dotted line ( only for @xmath20 and @xmath10 as sdss spectra do not cover the @xmath13 filter ) .
there is a small shift due to slight different definitions of the filter bandpasses ( also seen on other comparison data , rakos & schombert 2005b ) , but the slopes and dispersions are identical .
a strongly defined color locus opens the possibility of classifying galaxies by their spectrophotometric properties , in contrast to their morphological appearance .
while classification by morphological appearance has been a useful diagnostic for galaxy evolution , the underlying physics is difficult to unravel .
classification by optical color reflects , mostly , the physics of the star formation histories of galaxies , which is potentially more insightful to the evolution and formation of the stellar populations in galaxies . we have applied pca in our previous papers ( see odell , schombert & rakos 2002 ) to define four classes of galaxies based on their narrow band colors .
the most recent form of our classification system divides the first principal component axis ( pc1 ) into four subdivisions based on mean past star formation rate ; e ( passive , red objects ) , s ( star formation rates equivalent to a disk galaxy ) , s- ( transition between e and s ) and s+ ( starburst objects ) . by comparison to sed models and nearby galaxies (
rakos , maindl & schombert 1996 ) , the divisions along the pc1 axis are drawn such that s galaxies correspond to those systems with spiral star formation rates ( approximately 1 @xmath35 per yr ) and s+ galaxies correspond to starburst rates ( approximately 10 @xmath35 per yr ) .
we note that since these divisions are determined by continuum colors , versus spectral lines , they do not represent the current star formation rate , but rather the mean star formation rate averaged of the last few gyrs as reflected into the optical emission by the dominant stellar population .
the red e systems display colors with no evidence of star formation in the last five gyrs .
the transition objects , s- , represents the fact that there is not sharp division between the e class and s class .
these objects display slightly bluer colors ( statistically ) from the passive e class ; however , the difference could be due to a recent , low - level burst of star formation or a later epoch of galaxy formation or an extended phase of early star formation or even lower mean metallicity ( i.e. the color - magnitude effect ) .
in addition to classification by color , we can separate out objects with signatures of non - thermal continuum ( agn ) under the categories of a+ , a and a- based on their pc2 values colors .
it is important to remember that these classifications are based solely on the principal components as given by the color indices from four filters . while , in general , these spectrophotometric classes map into morphological ones ( i.e. e types are often ellipticals , s- are s0 and early - type spirals , s are late - type spirals and s+ are irregulars ) , this system differs from morphology by being independent of the appearance of the galaxy and based on the color of the dominant stellar population in a galaxy . this is also a classification based on integrated colors , such that large b / d galaxies ( i.e. early type spirals ) will generally be found in class s- as the bulge light dominates over the disk .
this is in contrast to classification by morphology where the existence of even a faint disk distinguishes the galaxy from an elliptical .
galaxies are labeled by their various spectrophotometric classes in figure 1 , where the size of the symbol is proportional to their absolute luminosity . unsurprisingly , a majority of the galaxies are e class ( 48% ) and s- ( 33% ) .
the remaining 19% were s class ( 15% ) and s+ ( 4% ) . the coma cluster ( odell , schombert & rakos 2002 ) has a similar mix of s / s+ galaxies at 12 and 5% each . however , in coma , a majority ( 72% ) of the red galaxies are e class and 14% are s- class , compared to 48% e class in a1185 versus 33% s- class .
it is important to remember that the s- class galaxies represents a transition set of colors between star - forming ( s class ) and non - star - forming ( e class ) as mapped into pc space . while s- class galaxies are , in general , associated with systems with low amounts of star formation ( large bulge disks , s0 s and e+a ) , they are also prominent as one examines lower luminosity systems where low mean metallicity pushes all the narrow band colors to the blue ( see figure 3 below ) . color ) as a function of absolute magnitude ( @xmath36 ) .
the drop in the fraction of e class galaxies as a function of luminosity reflects the color - magnitude relation for ellipticals rather any morphological effect .
the increase in star forming galaxies ( type s and s+ ) is real in that the blue population in a rich cluster increases with decreasing luminosity .
, width=642 ] the higher fraction of s- class galaxies in a1185 ( compared to coma ) is consistent with the higher fraction of blue galaxies found in a1185 .
the total blue fraction ( @xmath37 ) for a1185 , defined as the ratio of the number of galaxies with @xmath38 ( rakos 2000 ) is 0.21 ( trimmed to the same limiting absolute magnitude as coma ) . for comparison ,
the value for coma is 0.10 . given
that a1185 is a dynamically younger cluster than coma , a large blue fraction is to be expected ( see review of cluster environmental effects in poggianti 2004 ) .
a simple blue fraction value disguises a great deal of information about the population of galaxies in a cluster .
the bottom panel in figure 2 displays the run of @xmath37 as a function of absolute magnitude ( divided into bins of 32 galaxies ) . as with other rich clusters we have studied ( a2218 , a2125 rakos & schombert 2005b , a115 , a2283 rakos 2000 , a2317 rakos 1997 ) , a common trend of increasing @xmath37 with fainter magnitudes is found . while galaxies do become bluer with decreasing luminosity ( i.e. , cmr ) , this effect is minimal in the @xmath10 color where @xmath37 is defined ( see below ) .
the top panel in figure 2 also shows that fraction of star - forming galaxies ( s and s+ class ) also increases with decreasing luminosity and is the driving factor in @xmath37 .
one of the earliest photometric correlations for galaxies is the color - magnitude relation ( cmr ) and a1185 is unique by having the deepest examination of its cmr of any cluster of galaxies ( andreon 2006 ) .
the andreon study ( using @xmath39 colors ) found the cmr for a1185 to be linear and well defined down to absolute luminosities of @xmath40 .
figure 3 displays the cmr for all three narrow band colors in our study .
the solid lines are fits to the coma data and match perfectly to the a1185 , even at the level of the scatter around the relationships .
the universality of the cmr is well documented ( see schawinski 2006 and references therein ) , and its origin as a mass - metallicity relationship is also well - studied by theoretical models ( kodama & arimoto 1997 , although see trager 2000 for a dissenting view ) . while determining exactly how the age and metallicity of a galaxy map into the cmr requires guidance from sed models , it is known from our previous work in the coma cluster ( odell schombert & rakos 2002 ) that the primary driver of galaxy colors is metallicity , with age taking a secondary role if there is no recent star formation . the competition between age and metallicity is the main goal of this paper and we take the existence of the mass - metallicity relation for ellipticals to be given ( gallazzi 2006 ) . with respect to our narrow band colors ,
there are series of additional points .
first , the cmr is difficult to analyze as a linear relation due to the fact that any star - forming galaxies lie on the blue side of the relation which unbalances a linear fit .
our spectrophotometric classification system has the immediate advantage of removing star - forming galaxies from the fit , however , one then finds increasing weight given to red objects which biases the fit in the other direction .
in addition , the separation between e and s- classes is arbitrary ( although guided by morphological information ) , thus , final fits will be dependent on the cut used to separate the photometric categories .
fortunately , in a rich cluster , the number of red galaxies vastly outnumbers any contaminating blue galaxies .
experimentation with different cuts and bi - weight fitting procedures finds that the results for a1185 are statistically identical to cmr fits to other rich clusters in our previous papers .
the scatter on the final slope and zeropoint shown in figure 3 was due solely to observational uncertainty in the colors rather than sample criteria .
a second difference between the coma sample and a1185 is that the a1185 data involves a fainter magnitude cutoff .
we find that this resolves an anomaly in the coma @xmath10 cmr noted in odell , schombert & rakos ( 2002 ) .
for the coma data , we were surprised to find that the @xmath10 cmr had no slope in a formal fit .
our expectation from sed models was that a slope , due to metallicity effects , in @xmath28 and @xmath20 should translate into a smaller , but measurable slope in @xmath10 .
in @xmath20 , the cmr drops from a mean of 0.80 for @xmath41 to approximately 0.50 at @xmath42 . for a 13 gyr population ,
the same change in @xmath20 ( 0.30 ) corresponds to a change in 0.08 in @xmath10 , which was not seen in the coma data where a formal fit was equivalent to a zero slope . lacking a clear cmr in @xmath10 , we interpreted the lack of slope to be an age effect countering the mass - metallicity effect ( where @xmath10 is more sensitive to turnoff stars in an old stellar population and such that fainter galaxies were older ) .
the bottom panel in figure 3 indicates that the @xmath10 cmr for a1185 is , in fact , confused and a blend of galaxies at the faint end .
there is a clear correlation from the brightest galaxies to @xmath43 ( see dashed line fit in figure 3 ) , and the slope of this relation matches the expectations from the sed models and the @xmath20 slope .
but , below this luminosity , the relationship dissolves and the scatter increases such that a formal fit is a line of zero slope .
a similar effect is noted by poggianti ( 2001 ) for giant versus dwarf galaxies in coma .
also note , that due to this distortion in the @xmath10 cmr , our conclusions concerning the age of low luminosity galaxies in coma is invalid .
in fact , the poor resolution of the @xmath10 indicator to age , motivated us to seek a more intensive procedure to determine age and metallicity ( see 3.2 ) . which completely reverses our conclusions from odell , schombert & rakos ( 2002 ) concerning the age of low luminosity ellipticals . and
@xmath20 display the linear correlation for red galaxies that represents the run of metallicity with mass in ellipticals . while the coma data indicated a flat cmr for the @xmath10 index , the fainter data shown here for a1185 displays a normal cmr until @xmath43 then an upward trend is visible which distorts the linear fit .
the dashed line indicates a linear fit to the @xmath10 data for red galaxies brighter than @xmath44 .
, width=642 ] the data for a1185 also indicates there exists a spectrophotometric - density relationship in a1185 , equivalent to the density - morphology relation ( dressler 1984 , smith 1997 ) .
figure 4 displays the variation on projected galaxy density as a function of cluster radius for the four photometric classes ( s and s+ were combined to represent all star - forming galaxies ) .
unsurprisingly , star - forming galaxies avoid the cluster core and the red population ( e and s- classes ) dominate the inner regions .
numerous environmental effects have been postulated to account for rapid changes in the star formation of disk galaxies as they encounter a cluster core such as tidal forces , gas stripping , strangulation ( see poggianti 2004 for a review ) . however , even the s- class displays a statistically significant drop in density at the cluster core , even though there are no overt signs of star formation in these systems . an environmental effect on galaxies with old stellar populations would imply that the cluster environment has had a long history on galaxy populations , and that even within the red population of a cluster there are color differences influenced by environmental factors .
the goal for spectrophotometry studies of cluster galaxies is to reveal some information pertaining to their star formation history .
ultimately , we wish to determine not only the formation epoch of the galaxies ( the age of their oldest stars ) , but also the style and behavior of their star formation ( for example , initial bursts versus later episodes ) as it factors into their evolution .
the optical colors of galaxies reflect information that pertains to the photospheres of stars and , therefore , the evolution of a galaxy s colors mirrors changes in those same stellar populations .
thus , the light from stellar populations in galaxies is primarily a two parameter family based on the distribution of age and metallicity of the stars .
techniques for estimating the age and metallicity of galaxies have improved dramatically in the last 20 years . for optical studies ,
the primary technique of choice is age and metallicity determination through the use of spectral features ( gonzalez 1993 ) .
this has the immediate advantage of linking galaxy measurements to stellar studies , where a solid foundation of theoretical work on stellar atmospheres has resolved the behavior of spectral lines with temperature and abundance .
however , the dominant component that produces the observed colors in galaxies is not their spectral features but rather the continuum emission . age and metallicity vary the continuum emission through the impact of higher mass stars in younger stellar populations ( hot colors ) and changes in the color of the red giant branch ( rgb ) by line blanketing effects caused by changes in the mean metallicity ( tinsley 1980 ) .
stellar population studies by colors is not without drawbacks .
for example , the well known degeneracy problem for continuum colors derives from the fact that changes in metallicity can imitate age effects , particularly in observations without sufficient spectral resolution ( i.e. from broadband colors such as johnson @xmath45 ) .
our use of narrow band continuum colors is motivated by two factors ; 1 ) narrow filters select smaller regions of a galaxy s spectra , regions which contain a concentration of metal lines ( the @xmath11 region ) versus regions that are relatively free of line ( continuum areas , i.e. @xmath7 and @xmath8 ) and 2 ) there exist improved population models ( schulz 2002 ) which sample the region of age - metallicity parameter space useful for the analysis of old , red galaxies and which are now tied to the globular clusters system .
the schulz models are a significant improvement over past work with respect to our near - uv and blue colors since they 1 ) include the recent isochrones from the padova group which have a stronger blue hb contribution , 2 ) use more recent stellar atmosphere spectra which , again , are refined to better sample the blue and 3 ) are in better agreement with globular cluster @xmath4 colors than previous work . our studies on the use of the @xmath5 filters to determine age and metallicity have culminated in rakos & schombert ( 2005a ) where we develop a principal component methodology for determining these two parameters in composite systems , such as galaxies .
there are several limitations to the technique .
for example , the models are restricted to [ fe / h ] @xmath46 -1.7 and ages greater than 3 gyrs . however , for cluster galaxy studies , we can isolate galaxies with ongoing star formation , or have large fractions of young stars , by their extreme colors and/or strong emission lines .
in addition , there is a strong expectation that cluster galaxies consist of star populations with mean [ fe / h ] values greater than globular clusters , removing any problems of fitting the low metallicity end of the models .
our approach derives metallicity and age values using only the observed narrow band colors and the knowledge of the behavior of the pc surface ( i.e. , guided by the sed models ) .
we first use the three narrow band colors from the sed models ( schulz 2002 ) to generate a mesh of pc1 and pc2 values , where the pc equations ( see rakos & schombert 2005a ) contain age , metallicity and the three narrow band colors . with the expectation that one pair in the computed mesh represents the true value for an unknown age and metallicity , we next attempt to identify the true value of pc1-pc2 for an object with observed colors but unknown age and metallicity by an iterative search .
the search begins by using an estimated metallicity from the @xmath20 color and a crude , mean stellar age from the @xmath10 color .
we then interpolate a pc1 and pc2 pair from the assumed age and metallicity , use the pc equations to fill in the observed color indices plus the assumed age and metallicity , and compare the calculated pairs to those derived from the colors . in practice ,
the search for the correct age and metallicity is an iterative process as the pc plane is smooth , but not linear .
correct values are evaluated by calculating the differences between the pc values determined from models and those determined from the pc equations , in which the observed colors are the input .
the quality of the solution is measured by the root mean difference ( rms ) between the pc values over two iteration loops ; one for age and one for metallicity . at each step
, pc values are interpolated from the schulz et al .
( 2002 ) models for the step values of age and metallicity , then compared to the pc values determined from the pc equations , again using the step values of age and metallicity plus the observed colors .
there are several cautionary notes to this technique : 1 ) the procedure can only be used to determine mean ages and metallicities ( i.e. , objects which have multiple epochs of star formation will produce age/[fe / h ] values that are the luminosity weighted mean ) , 2 ) the technique only applies for stellar populations older than 3 gyrs ( i.e. , the degeneracy is still unresolved for galaxies undergoing starburst or recent star formation ) , 3 ) the age/[fe / h ] values are integrated values from the total luminosity of the galaxy and are not surface brightness selected ( e.g. core spectroscopy ) , this means that some metallicity / age model must be used to interpret internal stellar population distribution of age and metallicity as it reflects into the integrated colors .
the tools for our new age / metallicity algorithms can be found at http://student.fizika.org/@xmath47vkolbas/pca . .
there is a strong correlation between age and metallicity for the reddest galaxies ( bottom right panel ) with the oldest galaxies have the highest integrated [ fe / h ] values .
also plotted as solid lines are the hierarchical galaxy formation models of de lucia ( 2006 ) where the oldest galaxies were adjusted upward in age by 2.5 gyrs to match the ages of the oldest galaxies in our sample . while there is qualitative agreement between these simulations and the data , there are large numbers of systems that range in age and metallicity not predicted by the models . , width=642 ] the resulting age and [ fe / h ] values for the a1185 galaxies , as calculated using the technique described in rakos & schombert ( 2005a ) , are listed at our data website ( http://abyss.uoregon.edu/@xmath47js/narrow ) and displayed in figure 5 .
the typical error in our technique is 0.2 dex in [ fe / h ] and 0.5 gyrs in age which , of course , degrades at lower luminosities .
what is immediately obvious , in both the coma and a1185 data , is that cluster galaxies divide into two populations by age and [ fe / h ] .
the first population is a classic red sequence of galaxies with ages greater than 10 gyrs having a color - magnitude relation which is clear for high luminosity systems , but degrades at lower luminosities .
their morphological and spectrophotometric classifications are early - type ellipticals and s0 s with no current or recent star formation . the metallicities display strong chemical evolution with values that range from [ fe / h]@xmath48 to slightly greater than solar ( see also odell , schombert & rakos 2002 ) .
the second group of cluster members is younger than the first group with mean ages between 8 and 10 gyrs ( although for some of the star - forming members of this group , this age will represent an upper limit ) .
their metallicities also reflect a weakly evolved history with mean [ fe / h ] values less than @xmath49 .
an advantage to our narrow band color scheme is that these [ fe / h ] values are tied to the sed models .
as there are significant changes in @xmath2/fe ratios with galaxy mass ( sanchez - blazquez 2006 ) , [ fe / h ] extracted from line values must correct for this effect .
comparison to super - rich metallicity models ( rakos 2001 ) shows that changes in mg / fe from 0 to 0.4 result in changes in @xmath20 color of less than 0.01 mags .
the old , red group displays a classic trend of luminosity and [ fe / h ] for galaxies brighter than @xmath3218 , but the relation degrades at lower luminosities .
assuming that luminosity traces stellar mass ( cappellari 2006 ) , then bright red galaxies follow a well - defined mass - metallicity relation ( kelson 2006 , nelan 2005 ) .
comparison of all three narrow band colors also demonstrates that the cmr , in our color system , is primarily a metallicity indicator .
the effects of age are secondary , although in the same direction as metallicity effects with galaxy luminosity .
the low scatter in the cmr for red galaxies is not due to competing age and metallicity changes .
this older population also displays a clear correlation between age and luminosity ( stellar mass ) in the sense that more massive galaxies are older .
this trend indicates a mean age for high mass ellipticals to be 12.5 gyrs , on average , dropping to 11.5 gyrs for the faintest ellipticals in the sample .
this effect is often referred to as the ` downsizing ' scenario of galaxy formation ( smail 1998 , bundy 2005 , mateus 2006 ) and has been confirmed in line indices work ( collobert 2006 , gallazzi 2006 , sanchez - blazquez 2006 , gallazzi 2005 , thomas 2005 , kuntschner 2001 ) , optical colors ( cool 2006 ) , near - ir colors ( james 2006 ) , uv colors ( kaviraj 2006 ) , @xmath50 estimates ( di serego alighieri , lanzoni & jorgensen 2006 ) , field galaxies ( li , zhang & han 2006 , mateus 2006 ) , cluster galaxies ( nelan 2005 , poggianti 2004 , poggianti 2001 ) and galaxies at higher redshifts ( kelson 2006 , mei 2006 , schiavon 2006 , barr 2005 , jorgensen 2005 ) .
the best demonstrations of the downsizing effect are the cmr diagrams from @xmath51 clusters by de lucia ( 2004 ) .
these diagrams show a clear deficiency of low luminosity red galaxies between @xmath52 to @xmath44 .
our studies of the butcher - oemler effect in clusters , using the same narrow band filter set ( rakos & schombert 2005b ) , finds that only 5 to 10% of a galaxy s mass needs to be involved in star formation to produce spiral - like colors as seen in the de lucia diagrams .
given that @xmath51 corresponds to 6 gyrs ago ( benchmark cosmology ) , then this would produce a current generation of low luminosity cluster ellipticals with approximately 10% of their stellar mass in a 6 gyr population plus the remaining 90% of their stellar mass in a 12 gyr population .
convolving these values to our sed models finds that the luminosity weighted value for the age of such a galaxy would be 11 gyrs , in agreement with the age estimates from figure 5 .
the second group of galaxies , with younger mean ages , is represented in a1185 by two types of galaxies , a disk / irregular galaxy mixture on the low mass side combined with a few s- class objects on the high mass side . in coma ,
the younger population is primarily red , dwarf elliptical type systems ( see odell , schombert & rakos 2002 for more discussion ) .
the low mass red galaxies in younger group are primarily non - nucleated dwarf ellipticals , which we have found be have many of the characteristics of the blue globular clusters in ellipticals ( i.e. slightly younger and metal - rich compared to galactic globulars , rakos & schombert 2004 ) .
this supports our claim in rakos & schombert ( 2004 ) that low mass ellipticals divide into two populations , nucleated dwarfs with ages and metallicities that form a continuum with bright ellipticals and a younger non - nucleated dwarf elliptical population with slightly younger ages and higher metallicities .
our interpretation of blue galaxies in this second group , given their blue colors indicating ongoing star formation , is a population with a majority of their stellar mass in an old stellar population and a smaller mass , young ( less than 1 gyr ) stellar population which combine to produce an integrated age of 8 to 9 gyrs .
again , invoking estimates using sed models , we find that about 10% of the mass of these blue galaxies can be in a one gyr population and the remaining in a 12 gyr to produce the observed colors and ages .
it is interesting to note that the high mass systems in the second group appear to have the oldest ages , reddest colors and perhaps represent a transition population from the blue to red cluster galaxies ( possible future s0 s ) .
this population maintains many of the characteristics for the ` blue - side ' population found in sdss cmr studies ( cool 2006 , kauffmann 2003 ) , galaxies with primarily red stellar populations and a small percentage of stars with ages less than one gyr resulting in a dichotomy in color space by galaxy mass .
other points to note from figure 5 , the correlation between luminosity and metallicity for the red population is weaker than the correlation between mean age and metallicity . while a mass ( luminosity)-metallicity correlation is evident in the data , in fact , for values of [ fe / h ] above @xmath53 , the metallicity of a red galaxy so well correlated with age that either parameter predicts the other . the correlation between age and metallicity , in a positive sense , is in contradiction with other studies such as thomas ( 2005 ) and sanchez - blazquez ( 2006 ) which find anti - correlations between age and metallicity ( yet , positive correlations between galaxy mass and age or metallicity )
. correlated errors between the determination of age and metallicity could lead to positive correlation seen in figure 5 since redder colors can signify older age or higher metallicities .
however , examination of the least square fits to the parameters in pc space ( see figure 4 of rakos & schombert 2005 ) finds that @xmath54/@xmath55 has a slope of @xmath320.03 . whereas , the slope of the age / metallicity relation in figure 5 is @xmath560.4 .
although the range in age is low , it is unlikely that the correlation found from narrow band colors is due to coupling of the two parameters .
thus , figure 5 demonstrates that the range of age and metallicity vary among galaxies , particularly at low masses .
but for the oldest , high mass ellipticals , age and metallicity have a strong linear behavior such that [ fe / h ] = 0.36 @xmath57 - 4.55 in gyrs .
the existence of this tight relationship for the most massive galaxies is probably the strongest signature of a common origin and evolutionary path for these systems .
although the expectation from simple galactic wind models ( arimoto & yoshii 1987 , matteucci 1994 ) are that the oldest galaxies should have the lowest metallicities ( older meaning shortest duration of star formation and less processing of the ism ) , the reverse as seen in figure 5 .
this will be discussed further in the conclusions section .
a comparison between our spectrophotometric classifications and galaxy age is found in figure 6 .
the two age groups are clearly visible in this normalized histogram with mean ages at 12 and 8.5 gyrs . also noticeable in this histogram
is that even within the two groups there are differences in ages by spectrophotometric type .
the median age drops from 12 gyrs for e class galaxies to 11.5 gyrs for s- class galaxies to 11 gyrs for star - forming s / s+ class galaxies . for the later type galaxies ,
this probably reflects a mean age value based on a mixture of an old ( @xmath46 11 gyrs ) , bulge stellar population plus a younger disk population .
however , for the non - starforming s- class galaxies , their colors are typically more uniform and we would not propose that the younger age is due to a mixing of disk and bulge populations . yet , their ages from integrated light measurements are statistically younger than the e class galaxies .
the second , younger group is composed solely of s- , s and s+ class galaxies . where the youngest galaxies are , not surprisingly , the star - forming s and s+ galaxies .
however , the distribution in age for s- galaxies is distinctly bimodal with the youngest s- galaxies being older as the s / s+ galaxies ( 9.5 to 8.5 gyrs ) , but certainly separate from the peak at 11.5 gyrs .
the meaning of the trend with age and stellar mass is more difficult to extract from our data because the integrated age of the underlying population can be reached by many avenues .
for example , if all the stars in a galaxy are formed simultaneously , then our age measurement represents the time to that epoch .
this is also true if galaxies form their initial stars over a range of time , but all starting and stopping at the same epoch .
then our age measurement will represent the luminosity weighted mean age of the stars , presumably near the time to the median of the initial star formation epoch .
this enters into a debate of the formation time of the stellar population versus the assembly time of those populations into their parent galaxy ( de lucia 2006 ) . under a monolithic collapse scenario ,
those times are nearly equal .
but , recently , a variety of hierarchical models have been considered where stars forming in smaller units , then merger into their final systems .
one such set of models is presented by de lucia ( 2006 ) using the results of a high resolution cosmic structure growth simulation ( the millennium simulation , springel 2005 ) .
their resulting tracks of age , metallicity and galaxy mass ( converted to @xmath36 using an @xmath50 of 4 ) are shown in figure 5 .
the error bars on the models display the range from their simulations , not inaccuracies to their method . while there is qualitative agreement with the simulations in that age increases with galaxy mass and metallicity increases with mass
, the simulations appear incomplete at the low mass end and even the broad range of age in their low mass galaxy simulations do not capture the distribution of ages and metallicities on our sample . in particular , this flavor of models fails to find old , yet metal poor galaxies that represent a majority of the galaxies fainter than @xmath3217 . as an aside , an alternative interpretation of the luminosity - weighted ages arises if the duration of the epoch of initial star formation varies with galaxy mass .
for example , the red population trend in figure 5 may suggest that the initial phase of star formation is shorter for high mass galaxies than low mass ones .
an extended phase of initial star formation results in a mean age that is younger than the age calculated from a short , intense burst .
thus , the red population in a1185 and coma either 1 ) have low mass galaxies to be younger with respect to their initial star formation epoch or 2 ) have similar formation epochs , but low mass galaxies experience a longer duration of initial star formation as compared to higher mass systems .
this later interpretation is supported by the @xmath2/fe ratios in ellipticals ( denicolo 2005 , sanchez - blazquez 2006 ) , which increase with increasing galaxy mass to indicate a shorter duration of star formation .
evidence for recent star formation in elliptical ( barger 1996 ) would complicate the interpretation of our ages calculated from integrated colors .
these claims are based on spectroscopic signatures in the balmer series ( i.e. h@xmath58 ) . in an old population ,
very small numbers of hot , young stars are easy to detect in spectra . estimating the mass of a galaxy involved in a recent burst has not been conclusive .
our estimates from sed models is that a recent burst , which involves less than 5% of the galaxy mass , will effect on our narrow band colors at less than the level of the internal errors ( rakos & schombert 2005a ) .
thus , there is no conflict between observation of recent , weak bursts of star formation and our age estimates which will apply to a majority of the underlying stellar population . our coma sample ( odell , schombert & rakos 2002 ) lacked sufficient sky coverage to confidently address correlations related to distance from cluster center
. a larger field of view for the a1185 observations allows us to address those issues in this study .
figures 7 and 8 display just that information for a1185 , age and metallicity as a function of distance from the geometric center of the cluster .
there are weak , but statistically significant , trends for both parameters in the sense of decreasing metallicity and decreasing age with increasing distance from the cluster core .
these gradients are identical in direction and similar in magnitude to the line index gradients found in the noao fundamental plane survey ( smith 2006 ) .
they also found a decrease in age by 1 gyrs ( 15% ) and almost no metallicity gradient ( 2% ) .
their interpretation is later accretion produces younger stellar populations in a cluster halo , but early environmental effects , such as quenching , could equally well explain the observed gradients ( poggianti 2004 , bundy 2005 ) .
as we have already seen in previous section , there is a correlation between galaxy mass ( luminosity ) and age / metallicity .
so the trend with distance from the cluster core may simple reflect mass segregation . however , an examination of the galaxy luminosities as a function of cluster radius finds no correlation between galaxy luminosity and radial distance .
the change in mean stellar age is approximately 1 gyr from the cluster core to the outer regions .
this would represent a change in 2 to 3 mags in average galaxy luminosity as given by figure 7 , which is simply not seen in the data .
thus , the trends of age and metallicity are not due to changes in galaxy mass with cluster position . assuming that the trend of age with distance from the cluster core is real , and
not reflecting some secondary correlation with a galaxy s properties , we are surprised to find that metallicity - radius relation is so poor compare to the age - radius trend . for the red population in a1185 ,
the relationship between age and [ fe / h ] is very tight ( see 3.2 ) .
an age gradient of 1 gyr should be matched by a corresponding metallicity gradient of 0.5 dex , which is not seen in figure 8 .
this would agree with line strength studies ( bernardi 2006 ) which also found little change in fe indicators across various galaxy density regions .
while low metallicity systems appear to avoid the cluster core , high metallicity galaxies ( [ fe / h]@xmath59 ) are found at all cluster radii .
the implication here is that cluster environmental effects , which alter the red populations star formation history , affect the star formation rate ( i.e. age ) more sharply than the chemical evolutionary history of a galaxy , i.e. the mean metallicity of a red galaxy is set long before the era of star formation is complete . ) with cluster distance . as with figure 7
, there is a trend of decreasing galaxy metallicity with cluster distance .
the dashed line is from the hierarchical galaxy formation models of de lucia ( 2006 ) , which predict a small metallicity gradient in rich clusters .
, width=642 ] various hierarchical models of galaxy formation predict differences between the ages of galaxies formed in high density versus low density regions ( baugh , cole & frenk 1996 ) .
present - day clusters form in high density peaks of the primordial density field .
this leads to early collapse times for galaxies forming in these regions and , therefore , an older stellar population compared to field galaxies .
in addition , there exist simulations that display age and metallicity gradients within the cluster environment based on the hierarchical recipes ( de lucia 2006 , shown as dashed lines in figures 7 and 8) .
there is some evidence for this effect in spectroscopic studies ( bernardi 2006 , collobert 2006 ) ; however , the data presented herein does not sample into the field population as would be required to test this observation .
the age gradients seen in figure 7 follow the change in current galaxy projected density from core to cluster halo , but these outer density values are still a factor of three greater than the field , and there is no guarantee that these galaxies are occupying the same local density today as at their formation .
gradients found in simulations ( de lucia 2006 ) are primarily driven by mass segregation such that the distance to the center of a cluster is correlated loosely with time and , thus , is reflected in the mass correlated properties of the individual galaxies .
the processes , both internal and external , that influence the star formation history of cluster galaxies are so complex and interconnected , that even the determination of relatively accurate values for mean age and metallicity will not uniquely map into a single galaxy formation scenario due to the large number of possible paths a stellar population may take to end up with the integrated values that we measure . the best we can do is isolate the pure observables in this study and list the most likely candidates for the processes that a galaxy might take to acquire its properties
. this will , in the end , be a list of the most likely star formation scenarios combined with the known environmental influences on cluster galaxies . with respect to the red population in rich clusters ,
our work herein has found clear trends in galaxy age and metallicity with respect to stellar mass and environment .
the most massive galaxies in a cluster are the oldest with luminosity weighted ages of 12 to 13 gyrs .
these galaxies also have the highest metallicities , near solar . with decreasing stellar mass ( luminosity ) , the mean age decreases to approximately 11 gyrs .
likewise , the mean metallicity decreases to [ fe / h]=@xmath53 .
age and metallicity are strongly correlated , such that older mean age correlates to higher mean metallicities .
in addition to trends with galaxy mass , the mean age of the red population also changes with environment with the oldest galaxies found in the dense inner core and the youngest galaxies at the lower density cluster halos , although there is little change in global metallicity with cluster location . as stated in 3.2 ,
it is impossible to determine strictly from integrated values whether the variation in mean age with galaxy mass is due to varying epochs of initial star formation or varying durations .
however , an additional stellar clock is available to resolve this dilemma , the ratio of @xmath2 elements to fe .
the ratio of @xmath2 elements to fe is a function of the number of type ii versus type ia supernova in a galaxy s past .
type ia sn produce extra amounts of fe and decrease the ratio .
however , type ia sn require at least a gyr of time to build - up within a galaxy and , therefore , high fractions of @xmath2/fe indicate shorter duration times for star formation .
sdss ( sanchez - blazquez 2006 ) and deep line indices studies ( nelan 2005 ) have already demonstrated that @xmath2/fe increases with galaxy mass ( although see matteucci & chiappini 2005 for a dissenting view on the @xmath2 abundance problem ) .
the relationship of age with galaxy mass seen herein is consistent with a single epoch of galaxy formation and increasing initial durations of star formation with decreasing stellar mass .
this appears to be a weaker version of the so - called ` downsizing ' effect discovered in other age studies ( gallazzi 2006 ) since , by comparison to sed models , if as much as 10% the stellar mass of a low luminosity elliptical is involved in star formation at @xmath51 ( de lucia 2004 ) then this would produce luminosity - weighted mean ages of around 11 gyrs at the present epoch .
the homogeneous properties of the red population , for example , the tight cmr found in all rich clusters , has always argued strongly for a monolithic collapse scenario ( partridge & peebles 1967 , larson 1975 ) .
yet , the decrease in age with galaxy density ( radial changes ) found in this study is precisely the effect predicted by most hierarchical formation models ( yee 2005 , conselice 2006 ) and , more importantly , an increase in metallicity with galaxy mass in conjunction with a decreasing duration of initial star formation with galaxy mass is in direct contradiction of the classic galactic wind model of galaxy formation . under a normal wind model ( see matteucci 2004 ) , the efficiency of star formation is a decreasing function of galactic mass .
thus , the timescale of star formation is based on the dynamical free fall time .
larger mass systems have longer dynamical timescales and , therefore , smaller efficiencies and longer durations of initial star formation before the onset of a galactic wind halts the process .
this allows more time for chemical evolution to increase the abundance of metals and resulting in higher global metallicities for higher mass galaxies ( i.e. , the color - magnitude relation ) .
however , the @xmath2/fe ratios and the age - mass relation ( figure 5 ) argue exactly the opposite of the classic wind model , that star formation was shorter and more efficient with increasing galaxy mass . in order to maintain a homogeneous collapse scenario ,
one is required to propose a process where ellipticals form by merging protoclouds and the merger process introduces higher densities / velocities resulting in more efficient , and more abundant , star formation ( i.e. , the inverse wind model , matteucci 1994 ) .
this would be a hybrid galaxy formation scenario , mergers early on then reducing by monolithic collapse to a final state .
gradients exist in a1185 , a dynamically young cluster .
but the gradients are more discernible in age , than in metallicity .
environmental effects are usually thought to primarily influence the star - forming population in clusters , we note that even the red population of s- class galaxies display a gradient with respect to cluster density and effect that most have applied at the earliest formation times . these two observations would indicate that gradients are suggestive of a ` quenching ' behavior for cluster galaxies ( poggianti 1999 , bundy 2005 ) that has been in place even as the cluster was formed at redshifts greater than 5 .
thus , the age and metallicity gradients found in a1185 appear to be the effect of environmental processes imprinted at early epochs onto the red population .
this type of behavior is visible in the latest hierarchical models produced by collisionless cosmological simulations ( de lucia 2006 ) .
these models give a qualitative fit to our observations ( see model tracks in figure 5 ) if the oldest galaxies in the models have their ages increased by 2.5 gyrs .
key to those models is that the star formation epoch matches formation redshifts of 5 , but that galaxies assemble by mergers at redshifts below 2 .
our data can not address the assembly timescale for the red population , which is an issue to be investigated by space imaging .
the second population in a1185 and coma ( with mean ages around 9 gyrs , see figure 5 ) is associated with the blue population in clusters and , for higher redshifts , the butcher - oemler effect .
while most of this population has a mean age based on the luminosity - weighted sum of a small fraction of young stars plus a bulk of older ( 12 gyrs ) stars , a significant fraction of this group have non - star - forming colors ( class s- ) and probably represent transition galaxies between the butcher - oemler population at high redshifts and their descendants at the current epoch .
this blue population is also a candidate for the source of the metal - rich globulars found in many bright ellipticals . in rakos & schombert ( 2004 ) ,
we re - analyzed the narrow band colors of m87 globulars from jordan ( 2002 ) and found that the bimodal aspect of the m87 globulars was due to a old , metal - poor population ( blue globulars ) plus a younger , metal - rich population ( red globulars ) , in agreement with many other studies of extragalactic globulars ( see brodie & strader for a review ) .
we estimated the age of the red population of globulars to be between 3 and 4 gyrs younger than the blue globulars .
interestingly , this is similar to the difference in age between the old cluster galaxies and the blue population in a1185 and coma .
it suggests that , under a merger scenario for globular clusters in ellipticals ( ashman & zepf 1992 ) , the younger galaxies in clusters can be a source for the younger globulars in ellipticals .
financial support from austrian fonds zur foerderung der wissenschaftlichen forschung and nsf grant ast-0307508 is gratefully acknowledged .
this research has made use of the nasa / ipac extragalactic database ( ned ) , which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
we also wish to thank the faculty and staff of the university of arizona s steward observatory for the time allocated to us using 90prime on the bok telescope .
this material is based upon work supported by the national science foundation under grant no . 0307508 .
andreon , s. 2006 , , 369 , 969 arimoto , n. , & yoshii , y. 1987 , , 173 , 23 ashman , k. m. , & zepf , s. e. 1992 , , 384 , 50 barger , a. j. , aragon - salamanca , a. , ellis , r. s. , couch , w. j. , smail , i. , & sharples , r. m. 1996 , , 279 , 1 barr , j. , davies , r. , jrgensen , i. , bergmann , m. , & crampton , d. 2005 , , 130 , 445 baugh , c. m. , cole , s. , & frenk , c. s. 1996 , , 283 , 1361 bernardi , m. , et al .
2001 , arxiv astrophysics e - prints , arxiv : astro - ph/0110344 bernardi , m. , et al .
2003 , , 125 , 1882 bernardi , m. , nichol , r. c. , sheth , r. k. , miller , c. j. , & brinkmann , j. 2006 , , 131 , 1288 bregman , j. n. , temi , p. , & bregman , j. d. 2006 , arxiv astrophysics e - prints , arxiv : astro - ph/0604399 bundy , k. , ellis , r. s. , & conselice , c. j. 2005 , , 625 , 621 butcher , h. , & oemler , a. , jr .
1984 , , 285 , 426 cappellari , m. , et al.2006 , , 366 , 1126 collobert , m. , sarzi , m. , davies , r. l. , kuntschner , h. , & colless , m. 2006 , , 370 , 1213 conselice , c. j. 2006 , , 638 , 686 cool , r. j. , eisenstein , d. j. , johnston , d. , scranton , r. , brinkmann , j. , schneider , d. p. , & zehavi , i. 2006 , , 131 , 736 de lucia , g. , kauffmann , g. , & white , s. d. m. 2004 , , 349 , 1101 de lucia , g. , springel , v. , white , s. d. m. , croton , d. , & kauffmann , g. 2006 , , 366 , 499 denicol , g. , terlevich , r. , terlevich , e. , forbes , d. a. , & terlevich , a. 2005 , , 358 , 813 dressler , a. 1984 , , 22 , 185 fisher , d. , franx , m. , & illingworth , g. 1996 , , 459 , 110 gallazzi , a. , charlot , s. , brinchmann , j. , white , s. d. m. , & tremonti , c. a. 2005 , , 362 , 41 gallazzi , a. , charlot , s. , brinchmann , j. , & white , s. d. m. 2006 , , 370 , 1106 gonzlez , j. j. 1993 , ph.d .
thesis , gregg , m. d. , ferguson , h. c. , minniti , d. , tanvir , n. , & catchpole , r. 2004 , , 127 , 1441 huchtmeier , w. k. , sage , l. j. , & henkel , c. 1995 , , 300 , 675 james , p. a. , salaris , m. , davies , j. i. , phillipps , s. , & cassisi , s. 2006 , , 367 , 339 jimenez , r. , friaca , a. c. s. , dunlop , j. s. , terlevich , r. j. , peacock , j. a. , & nolan , l. a. 1999 , , 305 , l16 jordn , a. , ct , p. , west , m. j. , & marzke , r. o. 2002 , , 576 , l113 jrgensen , i. , bergmann , m. , davies , r. , barr , j. , takamiya , m. , & crampton , d. 2005 , , 129 , 1249 kaviraj , s. , rey , s. - . , rich , r. m. , lee , y. - . , yoon , s. - . , & yi , s. k. 2006 , arxiv astrophysics e - prints , arxiv : astro - ph/0601050 kauffmann , g. , heckman , t. , white , s. d. m. , charlot , s. , tremonti , c. , brinchmann , j. , bruzual , g. , peng , e. , seibert , m. , bernardi , m. , blanton , m. , brinkmann , j. , castander , f. , csbai , i. , fukugita , m. , ivezic , z. , munn , j. , nichol , r. , padmanabhan , n. , thakar , a. , weinberg , d. , york , d. 2003 , , 341 , 33 kelson , d. d. , illingworth , g. d. , franx , m. , & van dokkum , p. g. 2006 , arxiv astrophysics e - prints , arxiv : astro - ph/0606642 kodama , t. , & arimoto , n. 1997 , , 320 , 41 kodama , t. , arimoto , n. , barger , a. j. , & aragon - salamanca , a. 1998 , , 334 , 99 kuntschner , h. , lucey , j. r. , smith , r. j. , hudson , m. j. , & davies , r. l. 2001 , , 323 , 615 larson , r. b. 1975 , , 173 , 671 li , z. , zhang , f. , & han , z. 2006 , arxiv astrophysics e - prints , arxiv : astro - ph/0605610 mahdavi , a. , geller , m. j. , fabricant , d. g. , kurtz , m. j. , postman , m. , & mclean , b. 1996 , , 111 , 64 mateus , a. , sodr , l. , cid fernandes , r. , stasiska , g. , schoenell , w. , & gomes , j. m. 2006 , , 370 , 721 matteucci , f. 1994 , , 288 , 57 matteucci , f. 2004 , baryons in dark matter halos , matteucci , f. , & chiappini , c. 2005 , publications of the astronomical society of australia , 22 , 49 mei , s. , et al . 2006 , , 639 , 81 nelan , j. e. , smith , r. j. , hudson , m. j. , wegner , g. a. , lucey , j. r. , moore , s. a. w. , quinney , s. j. , & suntzeff , n. b. 2005 , , 632 , 137 odell , a. p. , schombert , j. , & rakos , k. 2002 , , 124 , 3061 partridge , r. b. , & peebles , p. j. e. 1967 , , 147 , 868 poggianti , b. 2004 , baryons in dark matter halos , poggianti , b. m. , smail , i. , dressler , a. , couch , w. j. , barger , a. j. , butcher , h. , ellis , r. s. , & oemler , a. j. 1999 , , 518 , 576 poggianti , b. m. , et al .
2001 , , 562 , 689 poggianti , b. m. , bridges , t. j. , komiyama , y. , yagi , m. , carter , d. , mobasher , b. , okamura , s. , & kashikawa , n. 2004 , , 601 , 197 rakos , k. d. , schombert , j. m. , & kreidl , t. j. 1991 , , 377 , 382 rakos , k. d. , & schombert , j. m. 1995 , , 439 , 47 rakos , k. d. , maindl , t. i. , & schombert , j. m. 1996 , , 466 , 122 rakos , k. d. , odell , a. p. , & schombert , j. m. 1997 , , 490 , 194 rakos , k. d. , schombert , j. m. , odell , a. p. , & steindling , s. 2000 , , 540 , 715 rakos , k. , schombert , j. , maitzen , h. m. , prugovecki , s. , & odell , a. 2001 , , 121 , 1974 rakos , k. , & schombert , j. 2004 , , 127 , 1502 rakos , k. , & schombert , j. 2005a , , 117 , 245 rakos , k. , & schombert , j. 2005b , , 130 , 1002 renzini , a. 2006 , arxiv astrophysics e - prints , arxiv : astro - ph/0603479 sanchez - blazquez , p. , gorgas , j. , cardiel , n. , & gonzalez , j. j. 2006 , arxiv astrophysics e - prints , arxiv : astro - ph/0604568 schawinski , k. , et al .
2006 , arxiv astrophysics e - prints , arxiv : astro - ph/0601036 schiavon , r. p. , et al .
2006 , arxiv astrophysics e - prints , arxiv : astro - ph/0602248 schulz , j. , fritze - v .
alvensleben , u. , mller , c. s. , & fricke , k. j. 2002 , , 392 , 1 smail , i. , dressler , a. , couch , w. j. , ellis , r. s. , oemler , a. j. , butcher , h. , & sharples , r. m. 1998 , vizier online data catalog , 211 , 213 smith , r. j. , lucey , j. r. , steel , j. , & hudson , m. j. 1997 , , 291 , 461 smoli , v. , et al .
2006 , , 371 , 121 springel , v. , et al .
2005 , , 435 , 629 thomas , d. , maraston , c. , bender , r. , & mendes de oliveira , c. 2005 , , 621 , 673 tinsley , b. m. 1980 , fundamentals of cosmic physics , 5 , 287 trager , s. c. , faber , s. m. , worthey , g. , & gonzlez , j. j. 2000 , , 120 , 165 tremonti , c. a. , et al .
2004 , , 613 , 898 worthey , g. 1999 , asp conf .
ser . 192 : spectrophotometric dating of stars and galaxies , 192 , 283 worthey , g. , & ottaviani , d. l. 1997 , , 111 , 377 yee , h. k. c. , hsieh , b. c. , lin , h. , & gladders , m. d. 2005 , , 629 , l77 di serego alighieri , s. , et al .
2005 , , 442 , 125 van dokkum , p. g. , & stanford , s. a. 2003 , , 585 , 78 williams , g. g. , olszewski , e. , lesser , m. p. , & burge , j. h. 2004 , , 5492 , 787
|
we present age and metallicities determinations based on narrow band continuum colors for the galaxies in the rich clusters a1185 and coma . using a new technique to extract luminosity - weighted age and [ fe / h ] values for non - star - forming galaxies , we find that both clusters have two separate populations based on these parameters .
one population is old ( @xmath0 11 gyrs ) with a distinct mass - metallicity relation .
the second population is slightly younger ( @xmath1 9 gyrs ) with lower metallicities and lower stellar masses .
we find detectable correlations between age and galaxy mass in both populations such that older galaxies are more massive and have higher mean metallicities , confirming previous work with line indices for the same type of galaxies in other clusters ( kelson 2006 , thomas 2005 ) . given the previously discovered correlation between galaxy mass and @xmath2/fe abundance ( a measure of the duration of initial star formation , denicolo 2005 , sanchez - blazquez 2006 ) , we interpret our age - metallicity correlations to imply that cluster galaxies are coeval with varying durations for their initial bursts .
our results imply shorter durations for higher mass galaxies , in contradiction to the predictions of classic galactic wind models . since we also find a clear mass - metallicity relation for these galaxies
, then we conclude that star formation was more efficient for higher mass galaxies , a scenario described under the inverse wind models ( matteucci 1994 ) . with respect to cluster environmental effects
, we find there is a significant correlation between galaxy mean age and distance from the cluster center , such that older galaxies inhabit the core .
this relationship would nominally support hierarchical scenarios of galaxy formation ( younger age in lower density regions ) ; however , environmental effects probably have larger signature in the sample and present - day galaxies are remnants from an epoch of quenching of initial star formation , which would result in the same age gradients .
|
we consider the geometric flow equations @xmath3 and ask whether closed strictly convex @xmath4-dimensional surfaces @xmath5 in @xmath6 converge to round points or to spheres at @xmath7 .
the answer is affirmative for many normal velocities @xmath8 , including certain powers of the gauss curvature , @xmath0 , the mean curvature , @xmath1 , and the norm of the second fundamental form , @xmath2 . here , authors like b. andrews @xcite , o. schnrer @xcite , and f. schulze @xcite , use functions of the principal curvatures @xmath9 to show convergence to a round point or to spheres at @xmath7 . in @xcite , @xcite
, o. schnrer proposes a characterization of these functions . and
in this paper , we extend it to non - rational functions .
our definition of _ maximum - principle functions _ ( mpf ) now covers any such function @xmath9 that is known to far .
the function from @xcite @xmath10 is an example of a mpf for a normal velocity @xmath11 with @xmath12 .
our main question is when mpf cease to exist for the large set of contracting and expanding normal velocities @xmath13 .
we are particularly interested in normal velocities @xmath14 , @xmath15 , and @xmath16 , for all powers @xmath17 . here , we either are able to prove the non - existence of mpf , or give an example of a mpf .
we present mpf from literature , and newly discovered mpf .
small gaps only exist for @xmath18 with @xmath19 , and @xmath20 with @xmath21 .
we summarize our main results in the table for @xmath13 ( p. ) , and in the tables for @xmath22 ( p. ) , for @xmath23 ( p. ) , for @xmath24 ( p. ) , and for @xmath25 ( p. ) .
+ _ the paper is structured as follows : _ in the chapter on notation , we give a brief introduction to differential geometric quantities like induced metric , second fundamental form , and principal curvatures .
the next chapter is on @xmath13 .
this is the set of normal velocities for which we investigate the existence and non - existence of mpf .
we proceed with a chapter on the mpf , motivating the definition and taking a closer look at the linear operator @xmath26 and the @xmath27-conditions .
chapter 5 contains main theorem [ thm : necessary ] . using euler s theorem on homogeneous functions we are also able to prove the non - existence of mpf for any normal velocity with @xmath28 . in chapter 6
we present vanishing functions , which are sometimes mpf , depending on the given normal velocity @xmath13 .
furthermore , we present a few technical lemmas on vanishing functions , which play an important role in the following chapters on @xmath22 , @xmath23 , @xmath24 , and @xmath25 .
we conclude our paper with an outlook suggesting an improved mpf ansatz .
for a brief introduction of the standard notation we adopt the a corresponding chapter from @xcite .
we use @xmath29 to denote the embedding vector of a @xmath4-manifold @xmath5 into @xmath6 and @xmath30 for its total time derivative .
it is convenient to identify @xmath5 and its embedding in @xmath6 .
the normal velocity @xmath8 is a homogeneous symmetric function of the principal curvatures .
we choose @xmath31 to be the outer unit normal vector to @xmath5 .
the embedding induces a metric @xmath32 and the second fundamental form @xmath33 for all @xmath34 .
we write indices preceded by commas to indicate differentiation with respect to space components , @xmath35 for all @xmath36 .
we use the einstein summation notation .
when an index variable appears twice in a single term it implies summation of that term over all the values of the index .
indices are raised and lowered with respect to the metric or its inverse @xmath37 , @xmath38 .
the principal curvatures @xmath39 are the eigenvalues of the second fundamental form @xmath40 with respect to the induced metric @xmath41 . a surface
is called strictly convex , if all principal curvatures are strictly positive .
we will assume this throughout the paper .
therefore , we may define the inverse of the second fundamental form denoted by @xmath42 .
symmetric functions of the principal curvatures are well - defined , we will use the gauss curvature @xmath43 , the mean curvature @xmath44 , the square of the norm of the second fundamental form @xmath45 , and the trace of powers of the second fundamental form @xmath46 .
we write indices preceded by semi - colons to indicate covariant differentiation with respect to the induced metric , e.g. @xmath47 , where @xmath48 .
it is often convenient to choose normal coordinates , coordinate systems such that at a point the metric tensor equals the kronecker delta , @xmath49 , in which @xmath40 is diagonal , @xmath50 .
whenever we use this notation , we will also assume that we have fixed such a coordinate system .
we will only use a euclidean metric for @xmath6 so that the indices of @xmath51 commute according to the codazzi - mainardi equations .
a normal velocity @xmath8 can be considered as a function of @xmath52 or @xmath53 .
we set @xmath54 , @xmath55 .
note that in coordinate systems with diagonal @xmath56 and @xmath49 as mentioned above , @xmath57 is diagonal .
in this chapter , we specify what we mean by a contracting and an expanding normal velocity and define the important quantity @xmath58 . in remark [ rem : normal velocity xi ] , we introduce the set of normal velocities @xmath13 .
this set includes powers of the gauss curvature , @xmath22 , powers of the mean curvature , @xmath23 , powers of the norm of the second fundamental form , @xmath24 , and the trace of powers of the second fundamental form , @xmath25 . throughout this paper ,
our goal is to determine , when mpf exist and cease to exist for @xmath13 .
[ def : normal velocity ] let @xmath59 and @xmath60 be principal curvatures .
let @xmath61 be a symmetric homogeneous function of degree @xmath17 . in this paper , we call @xmath8 a _ normal velocity _ if @xmath62 for all @xmath63 . furthermore , we call @xmath8 _ contracting _ if @xmath64 for all @xmath63 , and we call @xmath8 _ expanding _ if @xmath65 for all @xmath63 .
[ def : beta ] let @xmath8 be a normal velocity .
we define the quantity @xmath66 as @xmath67 for all @xmath63 .
we later choose @xmath68 and write @xmath69 .
[ rem : normal velocity xi ] in this paper , we investigate the normal velocity @xmath13 .
we define @xmath13 as @xmath70 for all @xmath17 . calculating @xmath71 , @xmath72
, we obtain @xmath73 and @xmath74 for all @xmath63 .
therefore , @xmath75 for all @xmath63 .
hence , @xmath13 is a normal velocity for all @xmath76 .
@xmath13 is a _ contracting _ normal velocity , if @xmath77 , and @xmath13 is an _ expanding _ normal velocity , if @xmath78 . calculating the quantity
@xmath79 we obtain @xmath80 for all @xmath76 .
note that @xmath81 is independent of @xmath82 .
let @xmath83 .
then we have @xmath84 , and @xmath85 let @xmath86 . then we have @xmath87 , and @xmath88 let @xmath89 . then we have @xmath90 , and @xmath91 let @xmath92 . then we have @xmath93 , and @xmath94
in the context of geometric evolution equations , @xmath95 the question arises , if after appropriate rescaling , closed strictly convex surfaces @xmath5 converge to spheres .
this is also referred to as convergence to a round point or convergence to a sphere at @xmath7 .
the answer is affirmative for many normal velocities @xmath8 , including the gauss curvature flow , @xmath96 , b. andrews @xcite , the mean curvature flow , @xmath97 , g. huisken @xcite , and the inverse gauss curvature flow , @xmath98 , o. schnrer @xcite . for many normal velocities @xmath8 proofs rely on
the fact that a certain geometrically meaningful quantity @xmath9 is monotone , @xmath99 is non - increasing in time . in @xcite and @xcite
, o. schnrer proposes criteria for selecting such monotone quantities for contracting flows and for expanding flows , respectively . to date , to the author s knowledge ,
all known quantities which fulfill these criteria can be used to show convergence to a round point or to a sphere at @xmath7 . here , we decided to work with these criteria as a definition .
their monotonicity is proven using the maximum - principle .
this is why we name these quantities _ rational maximum - principle functions _ ( rmpf ) .
[ def : rmpf ] let @xmath59 and @xmath60 be principal curvatures .
let @xmath100 be a symmetric homogeneous function of degree @xmath101 .
we call @xmath9 a _ rational maximum - principle function _ for a normal velocity @xmath8 if 1 .
[ ri ] 1 .
@xmath102 for all @xmath63 , @xmath103 , 2 .
@xmath104 for all @xmath105 .
[ rii ] 1 .
@xmath106 if @xmath8 is contracting , 2 .
@xmath107 if @xmath8 is expanding .
3 . [ riii ] 1 .
@xmath108 for all @xmath109 , 2 .
@xmath110 for all @xmath111 .
[ riv ] let @xmath112 be the linear operator corresponding to the geometric flow equation .
we achieve @xmath113 for all @xmath63 by assuming that 1 .
terms without derivatives of @xmath40 are non - positive for all @xmath63 , and 2 .
terms involving derivatives of @xmath40 , at a critical point of @xmath9 , _
@xmath114 for all @xmath115 , are non - positive for all @xmath63 .
[ rv ] @xmath116 is a rational function .
the rmpf conditions through as in @xcite , @xcite are motivated as follows : for all geometric flow equations we assume that spheres stay spherical . for contracting ( expanding ) normal velocities they contract to a point ( expand to infinity ) .
so we can only aim to find monotone quantities , if @xmath117 for all @xmath118 , or if @xmath119 ( if @xmath120 ) .
if @xmath119 ( if @xmath120 ) , we obtain that @xmath9 is non - increasing on any self - similarly contracting ( expanding ) surface .
so this does not imply convergence to a round point .
rmpf condition ensures that the quantity decreases if the principal curvatures approach each other . by rmpf condition
, we check that we can apply the maximum principle to prove monotonicity .
the first found monotone quantities are all rational functions , @xmath121 for @xmath96 in @xcite , or @xmath122 for @xmath123 in @xcite .
this motivates rmpf condition .
[ def : mpf ] let @xmath59 and @xmath60 be principal curvatures .
let @xmath100 be a symmetric homogeneous function of degree @xmath101 .
we call @xmath9 a _ maximum - principle function _ for a normal velocity @xmath8 if 1 .
[ i ] 1 .
@xmath102 for all @xmath63 , @xmath103 , 2 .
@xmath104 for all @xmath105 .
[ ii ] 1 .
@xmath106 if @xmath8 is contracting , 2 .
@xmath107 if @xmath8 is expanding .
[ iii ] 1 .
@xmath108 for all @xmath109 , 2 .
@xmath110 for all @xmath111 .
[ iv ] we assume that the _ constant terms _ @xmath124 and the _ gradient terms _ @xmath125 and @xmath126 are non - positive for all @xmath63 .
@xmath127 5 .
[ v ] we define @xmath128 let @xmath129 , @xmath130 be some constants . set @xmath68 , where @xmath131 .
if @xmath8 is contracting , we assume for @xmath132 @xmath133 if @xmath8 is expanding , we assume for @xmath132 @xmath134 f. schulze and o. schnrer present in @xcite one of the first non - rational monotone quantities .
they use @xmath10 to show convergence to a round point for @xmath135 , for all @xmath12 .
here , @xmath9 is not rational , if @xmath136 .
therefore , in mpf definition [ def : mpf ] , we extend rmpf definition [ def : rmpf ] to not necessarily rational functions .
we call them _ maximum - principle functions _ ( mpf ) . in lemma
[ lem : rmpf ] , we prove that mpf condition holds for any rmpf in the case of contracting flows @xmath13 , if @xmath137 , or @xmath83 and @xmath138 .
for other @xmath13 , a similar lemma is in progress .
a first step in the this direction is lemma [ lem : alpha o estimates ] . to date , all known rmpf fulfill mpf condition . in this paper , the main question is , when mpf cease to exist .
as it turns out , this is more an algebra than a differential geometry question .
correspondingly , in mpf condition we formulate three inequality conditions . in lemmas [ lem:1stderivatives ] through [ lem : lwcriticalpoint ] , we show that these three inequality conditions are equivalent to rmpf condition . in condensed form , mpf definition
[ def : mpf ] is a more algebraic formulation of rmpf definition [ def : rmpf ] , which is extended to what seems to be the proper class of non - rational functions . in lemmas [ lem:1stderivatives ] through [ lem : lwcriticalpoint ] , we show that the three inequality conditions @xmath139 in condition mpf are equivalent to rmpf condition as follows : in lemma [ lem:1stderivatives ] , we show two helpful identities for the first derivatives with respect to the induced metric @xmath140 , and for the derivatives with respect to the second fundamental form @xmath56 .
next , we cite the evolution equations for @xmath140 and @xmath56 from o. schnrer @xcite . we need both lemmas in lemma [ lem : lw ] . here , we compute the linear operator @xmath141 , where @xmath9 is a function of the principal curvatures @xmath59 and @xmath60 . sometimes , we also denote them by @xmath142 and @xmath143 . in lemma [ lem:2ndderivatives ] , we cite another helpful identity for the second derivatives with respect to the second fundamental form .
finally , in lemma [ lem : lwcriticalpoint ] we compute the constant terms @xmath124 , and the two gradient terms @xmath125 and @xmath126 .
this concludes the proof of our claim that condition of rmpf definition [ def : rmpf ] and of mpf definition [ def : mpf ] are equivalent .
[ lem:1stderivatives ] let @xmath144 be a normal velocity @xmath8 or a function @xmath9 of the principal curvatures @xmath142 and @xmath143 .
then we have @xmath145 the matrices @xmath146 and @xmath147 are symmetric .
we consider @xmath148 .
the matrices @xmath40 and @xmath41 are symmetric .
so we have @xmath149 and @xmath150 .
+ now , we differentiate @xmath144 with respect to @xmath151 .
@xmath152 since @xmath153 for all @xmath154 , the matrix @xmath146 is symmetric and formula follows .
+ next , we differentiate @xmath144 with respect to @xmath155 .
@xmath156 since @xmath157 for all @xmath158 the matrix @xmath147 is symmetric and formula follows .
[ lem : evolution ] let @xmath159 be a solution to a geometric flow equation
. then we have the following evolution equations : @xmath160 where @xmath26 is the linear operator of rmpf definition [ def : rmpf ] .
we refer to o. schnrer @xcite .
[ lem : lw ] let @xmath161 be a function of the principal curvatures
. then we have @xmath162 where @xmath26 is the linear operator of rmpf definition [ def : rmpf ] .
we consider @xmath163 .
@xmath164 [ lem:2ndderivatives ]
let @xmath144 be a normal velocity @xmath8 or a function @xmath9 of the principal curvatures @xmath142 and @xmath143 .
then we have @xmath165 for any symmetric matrix @xmath166 and @xmath167 , or @xmath168 and the last term is interpreted as a limit .
we refer to c. gerhardt @xcite .
[ lem : lwcriticalpoint ] let @xmath169 be a symmetric function of the principal curvatures @xmath59 and @xmath60 . at a critical point of @xmath9 , @xmath114 for all @xmath170
, we choose normal coordinates , @xmath171 and @xmath172 .
+ then we have @xmath173 where @xmath26 is the linear operator of rmpf definition [ def : rmpf ] , and @xmath27 is the quantity of mpf definition [ def : mpf ] . + the constant terms are @xmath174 and the two gradient terms are @xmath175 we obtain @xmath176 for all @xmath63 .
we consider @xmath177 . at a critical point of @xmath9 ,
we have @xmath178 here , we also choose normal coordinates and get @xmath179 , implying @xmath180 and @xmath181 , implying @xmath182 now , we compute the linear operator @xmath141 of lemma [ lem : lw ] at a critical point of @xmath9 , where we choose normal coordinates .
@xmath183 clearly , the constant terms @xmath184 are the first two lines of terms after the last equation equal sign .
it remains to compute the gradient terms using identities and .
we obtain the first gradient terms @xmath185 and we obtain second gradient terms @xmath186 rmpf definition [ def : rmpf ] and mpf definition [ def : mpf ] are the same up to condition , and in lemma [ lem : lwcriticalpoint ] we have shown that rmpf and mpf conditions are equivalent .
so it only remains to motivate mpf condition .
rmpf condition assumes any such function to be rational .
as mentioned before , the example of @xmath187 given by o. schnrer and f. schulze @xcite fulfills all rmpf conditions except rmpf condition for any @xmath188 .
but it can be used to show convergence to a round point using the maximum - principle . as a first step , we analyzed all functions that were used to show convergence to a round point and convergence to a sphere at @xmath7 in @xcite , @xcite , @xcite , @xcite , @xcite , and @xcite .
all of them have in common that they fulfill mpf condition . as a second step ,
we seek to prove that all rmpf are mpf .
this is lemma [ lem : rmpf ] . here
, we show that for contracting @xmath13 with @xmath137 , or @xmath83 and @xmath138 , rmpf fulfill mpf condition .
this includes the contracting normal velocities @xmath189 , @xmath135 , and @xmath190 .
a similar lemma for the remaining contracting and the expanding @xmath13 is in progress . in @xcite
, we show that there are no rmpf for any @xmath191 with @xmath192 . and due to b. andrews we know that convex surfaces do not necessarily converge to a round point for any power @xmath193 . for the power @xmath194
, they converge to ellipsoids @xcite . for any power @xmath195 , surfaces contract homothetically in the limit @xcite .
this is why we can not expect any rmpf to exist for any @xmath191 with @xmath196 .
+ we begin this chapter by showing that the quantity @xmath197 is strictly positive for all @xmath63 , if @xmath9 is a rmpf or a mpf . in remark [ rem :
examples condition ] , we give an example for each of the @xmath27-conditions in mpf condition
. then we proceed with lemma [ lem : alpha o estimates ] , which contains some calculations for lemma [ lem : rmpf ] . by this lemma
, we motivate mpf condition .
[ lem : alpha ] let @xmath9 be a rmpf or a mpf
. then we have @xmath198 for all @xmath199 .
note that @xmath197 as in mpf definition [ def : mpf ] .
let @xmath199 .
mpf condition and the symmetry of @xmath9 in @xmath59 and @xmath60 imply @xmath200 recall , @xmath197 .
hence , the claim follows for all @xmath63 with @xmath103 .
now , let @xmath63 with @xmath201 . by mpf condition , in particular @xmath202 for all @xmath63 , we have @xmath203 for all @xmath63 .
this implies @xmath204 for all @xmath63 . + letting @xmath109 yields @xmath205 , and letting @xmath111 yields @xmath206 . hence ,
for @xmath207 we get @xmath208 this implies @xmath209 for all @xmath63 with @xmath201 .
we get @xmath210 let @xmath211 , then @xmath212 since @xmath213 is homogeneous , we have @xmath214 for all @xmath63 with @xmath201 .
this concludes the proof .
[ rem : examples condition ] we give an example for each of the @xmath27-conditions in mpf condition .
let @xmath215 .
then we have [ cols="<,^,^,^,^",options="header " , ]
` the search is over . ' + at least , this is what had been our second choice for a title of this paper .
it sounded a bit too dramatic , so we decided to go for the first .
however , the mpf ansatz followed by b. andrews , o. schnrer , f. schulze , and many more , now seems to be exploited to the last drop . certainly that is for the gauss curvature , @xmath22 , the mean curvature , @xmath23 , and the norm of the second fundamental form , @xmath24 .
but the search is not over . in our opinion , mpf condition is sufficient , not necessary , and obviously far too restrictive for many normal velocities .
for example , in case of @xmath22 , we know that no mpf exists for any power @xmath216 , but we still have convergence to a sphere at @xmath7 by c. gerhardt @xcite . and there are no apparent reasons why convergence to a round point should stop for powers of the mean curvature greater than 6 , or for powers of the norm of the second fundamental form greater than 11 .
we conjecture that we can exploit the mpf ansatz further , if we can find a way to weaken mpf condition
. then we also might be able to recycle some of the vanishing functions that are not mpf according to the present mpf definition [ def : mpf ] .
|
we consider geometric flow equations for contracting and expanding normal velocities , including powers of the gauss curvature , @xmath0 , of the mean curvature , @xmath1 , and of the norm of the second fundamental form , @xmath2 , and ask whether - after appropriate rescaling - closed strictly convex surfaces converge to spheres . to prove this , many authors use certain functions of the principal curvatures , which we call maximum - principle functions .
we show when such functions cease to exist and exist , while presenting newly discovered maximum - principle functions .
|
the structure of the qcd phase diagram is intimately related to our understanding of fundamental features of qcd , like for example confinement dynamics and chiral symmetry breaking , and to their interplay with in - medium effects like a nonzero temperature or quark density .
this is the reason why a lot of effort is devoted to study this field , either on the theoretical side , to which the present work belongs , or on the experimental side through heavy - ion - collision experiments . among the various effective frameworks used to study the qcd phase diagram ( see _ e.g. _ the review @xcite ) , we will mostly focus on two of them : polyakov - loop effective models for the pure gauge part of qcd , and the nambu - jona - lasinio ( njl ) model for the quark part .
the polyakov loop is defined as @xmath8 in which @xmath9 is the path - ordering , @xmath10 the strong coupling constant , @xmath11 the temporal component of the yang - mills field , @xmath12 the generators of the gauge algebra , and @xmath2 the temperature .
the integral runs on the compactified timelike dimension .
the polyakov loop is a precious tool to study the phase structure of a given yang - mills theory since @xmath13 @xmath14 when the theory is in a ( de)confined phase @xcite .
moreover , gauge transformations belonging to the center of the gauge algebra only cause @xmath15 to be multiplied by an overall factor . that is why it has been conjectured that the confinement / deconfinement phase transition in a yang - mills theory with gauge algebra @xmath16 might be related to the spontaneous breaking of a global symmetry related to the center of @xmath16 @xcite . in the particular case of su(@xmath1 ) , deconfinement might thus be driven by the breaking of a global z@xmath0 symmetry .
the order parameter of the deconfinement phase transition should then be the traced polyakov loop @xmath17 where the trace @xmath18 is taken over the colour indices .
the thermodynamic properties of pure gauge su(3 ) qcd can then be studied by resorting to an effective scalar field theory where the potential energy density is z@xmath19-symmetric , with _
e.g. _ the form @xcite @xmath20.\ ] ] the real coefficients @xmath21 can be fitted on lattice data .
various applications of this formalism can be found for example in @xcite .
note that , in the following , @xmath22 and @xmath23 will generally be indifferently called polyakov loop .
the njl model is based on the lagrangian @xcite @xmath24,\ ] ] where @xmath25 is the quark field , @xmath26 the mass matrix , and @xmath27 the pauli matrices when an su(2 ) flavour symmetry is considered .
the interaction terms are such that the lagrangian is chirally symmetric .
the njl model is designed to model chiral symmetry breaking and study many related phenomenological problems ; the interested reader may consult the review @xcite for more information . in the original njl model ,
fermions are not coupled to the gauge field : as shown in @xcite , the coupling of this model to the polyakov loop can be achieved by minimally coupling the quark field to a gauge field of the form @xmath28 , that formally appears as an imaginary quark chemical potential .
the so - called pnjl model resulting in this coupling has motivated a lot of studies devoted to the qcd phase diagram @xcite , including cases with a nonzero magnetic field @xcite or nonlocal extensions @xcite . limit ( @xmath29 ) of the pnjl model used in @xcite .
the solid lines signal first - order phase transitions.,width=321 ] the phase structure of the pnjl model at arbitrary @xmath1 has been discussed in @xcite , as well as its large-@xmath1 limit .
one of the ingredients of this last work is to set @xmath30 in ( [ ud ] ) so that the gluon potential has the correct scaling in @xmath1 .
the phase diagram that has been found at large-@xmath1 is given in fig .
the chirally symmetric but confined phase that appears for quark chemical potentials larger than about a third of the nucleon mass , @xmath31 , can presumably be identified with the quarkyonic phase , that has been first proposed in @xcite and further studied in @xcite in particular . in the present work ,
we propose to re - build a pnjl model valid at large-@xmath1 , but in which the polyakov - loop potential is explicitly z@xmath0 symmetric . a possible way to answer to this requirement is to introduce a term in @xmath32 instead of the standard z@xmath19-symmetric term @xmath33 . such a potential is proposed in sec .
[ pureglu ] , and the corresponding pnjl model is written in sec .
[ pnjls ] .
then , the issue of deconfinement and chiral symmetry restoration when varying @xmath2 and @xmath4 are discussed in sec . [ ccs ] in t hooft s large-@xmath1 limit . the obtained phase diagram and concluding comments are given in sec .
[ conclu ] .
the simplest effective potential energy density depending on @xmath22 , defined in ( [ phi ] ) , and being explicitly z@xmath0-invariant has been proposed in @xcite and reads @xmath34 it is formally valid for any value of @xmath22 , but one may restrict oneself to @xmath35 $ ] in a mean - field approximation .
the above expression contains the basic blocks that could be expected to build a nontrivial theory : a mass term ( @xmath36 ) , an interaction term ( @xmath37 ) , and the term in @xmath32 accounting for the explicit z@xmath0-symmetry @xcite .
terms scaling like @xmath38 , @xmath39 , @xmath40 , @xmath41 , etc . could be added , but then the number of arbitrary functions would become too large to be efficiently constrained
. moreover , such higher - order terms would mostly be interaction terms that are already present in their simplest form in the @xmath37 term .
the expression ( [ v1 ] ) is thus particularly convenient since it contains the minimal number of terms needed to perform the present study .
the real coefficients @xmath42 , @xmath43 , and @xmath44 appearing in ( [ v1 ] ) are functions of @xmath2 and @xmath1 , and their explicit form will be specified in the following .
note that @xmath22 , which depends on @xmath2 , @xmath1 , and @xmath45 a priori , is here assumed to be independent of @xmath45 . beyond the polynomial form ( [ v1 ] )
, logarithmic shapes can actually be shown to emerge from a haar integration on the gauge group in a strong coupling expansion .
one can find such a form in @xcite , or for example in @xcite , where a potential schematically given by @xmath46 $ ] is used for @xmath7 computations . instead of computing a similar potential at arbitrary @xmath1
, we keep the ansatz ( [ v1 ] ) in the following ; it is indeed particularly convenient for the calculations that are to be performed and still contains the z@xmath0 symmetry we want to take into account .
various parametrizations of z@xmath19-symmetric potentials , fitted on pure gauge lattice data , have been proposed so far @xcite . here , we are rather interested in obtaining an effective potential valid at large @xmath1 , _
i.e. _ @xmath47 at least .
as shown below , all these values of @xmath1 will have in common that , in our approach , theasymptotic " behaviour of @xmath48 ( _ i.e. _ values of @xmath49 larger that the physical one minimizing the potential ) will be driven by the @xmath50-symmetric term
. the present formalism will thus not be valid for @xmath7 in particular , where this asymptotic behaviour is driven by the interaction term .
the following expected qualitative behaviours have to be imposed in order to constrain the shape of the functions @xmath42 , @xmath43 , and @xmath44 : * the pressure @xmath51 is proportional to @xmath52 at large @xmath1 and @xmath2 in order to recover asymptotically the stefan - boltzmann limit of a free gluon gas . *
the norm , @xmath53 , of the optimal value of the polyakov loop , @xmath54 , is @xmath1-independent at the dominant order , see the definition ( [ phi ] ) .
the first corrections , scaling as @xmath55 , are neglected in the present approach
more results on large-@xmath1 features of wilson and polyakov loops can be found for example in @xcite .
@xmath56 in the confined phase , and @xmath57 in the deconfined phase . also , @xmath53 tends toward unity at very large @xmath2 .
* there exists a critical temperature @xmath3 signalling a first - order phase transition , _
i.e. _ the potential must have two different minima whose depth changes with the temperature in order to modify discontinuously the localisation of the absolute minimum . at the critical temperature , @xmath56 and @xmath58
are two degenerate minima of @xmath48 .
this last value is chosen so that it will ensure a good compatibility between our model and existing lattice data but it has only to be nonzero in order to lead to a deconfined phase .
@xmath3 has to be seen as a typical value for the deconfinement temperature in su(@xmath1 ) yang - mills theory since the deconfinement temperature appears to be @xmath1-independent up to corrections in @xmath55 @xcite .
obviously , the above constraint does not apply to @xmath59 , where the transition is of second - order .
this is not problematic since we eventually look for a model valid at large-@xmath1 .
moreover , the value @xmath60 may not be the exact value of the polyakov loop in @xmath3 : recent lattice results find it to be around 0.4 @xcite , while a more recent renormalization - group - based approach leads to values closer to 0.6 for the polyakov loop at the deconfinement temperature @xcite .
the value 1/2 then appears to be relevant because it falls in the typical range of the existing results and because it simplifies the calculations performed in the following .
the above constraints are actually satisfied by the following lagrangian @xmath61(\phi^{n_c}+\phi^{*n_c})\right],\ ] ] where @xmath62 explicit forms of @xmath63 and @xmath64 will be given in the next section .
all these conditions are required in order to have the existence of 2 degenerate minima and the correct behaviour of the polyakov loop in the mean field approximation .
the potential ( [ v2 ] ) has the following absolute minimum : @xmath65 and @xmath66 , where @xmath67 and where @xmath68 is a solution of @xmath69 |\phi_0(t)|^{n_c-2}=0.\ ] ] it is straightforwardly checked that @xmath70 actually solves ( [ eq1 ] ) .
a more compact expression for the optimal value of the polyakov loop is thus @xmath71 where @xmath72 is the heaviside function . as seen from ( [ eq1 ] )
, @xmath53 only depends on @xmath2 as required . restricting ourselves to the values @xmath73 , we get at the limit @xmath29 a quite simple shape for the effective potential
( [ v2 ] ) , namely @xmath74 hence , a u(1 ) invariance is recovered at infinite @xmath1 as a limiting case of the z@xmath0-symmetry .
the schematic evolution of the large-@xmath1 limit of @xmath48 with the temperature is plotted in fig .
[ fig2 ] ; the behaviour ( [ v3 ] ) is readily observed , as well as the change of global minimum in @xmath75 . )
versus the temperature ( solid lines ) .
, width=321 ] finally , the large-@xmath1 limit of the pressure reads @xmath76.\ ] ] provided that @xmath77 according to the large-@xmath2 behaviour of the polyakov loop , @xmath78 would tend toward the stefan - boltzmann limit for a free gluon gas if @xmath79 .
the function @xmath64 is constrained by the relations ( [ v2c ] ) in order for the structure of the potential and its evolution with the temperature to have the required behaviour .
moreover , @xmath64 is equal to the norm of the polyakov loop as soon as @xmath80 .
those physical constraints are not sufficient to write down an explicit expression for @xmath64 .
a possible way of proceeding , that we choose here , is to fit @xmath64 on available lattice computations of the polyakov loop in pure yang - mills theory . to our knowledge
, large-@xmath1 values have not been obtained so far , but accurate su(3 ) ones have been computed in @xcite . since the polyakov loop should not depend on @xmath1 at the dominant order ,
it is relevant to fit @xmath64 on su(3 ) data ; the ad hoc form @xmath81\ ] ] leads to a satisfactory parametrization of the results of @xcite as it can be seen in fig .
it is also worth noting that fig .
[ fig2 ] has been obtained using the form ( [ ldef ] ) for @xmath64 .
it is important to remark at this stage that the calculations we will perform are done in the mean - field approximation . in this scheme ,
the polyakov loop is always lower than 1 : values larger than 1 are due to quantum fluctuations and are _ de facto _ beyond the mean - field treatment .
that is why we have restricted our fit to lattice data lower than unity ( @xmath82 2.4 @xmath3 ) .
we miss the overshoot due to quantum fluctuations , but we stay coherent with the mean - field approximation , and reach moreover @xmath77 . as a consequence , our results should be mostly trusted below @xmath82 2.4 @xmath3 but this is not a flaw since , in the following , we will be concerned with the phase structure of the theory and no phase transition will appear at energy scales above this upper limit . ) versus the temperature in units of @xmath3 ( solid line ) .
the function @xmath64 ( dashed line ) and the norm of the polyakov loop computed in pure gauge su(3 ) lattice qcd ( points ) have been added for comparison .
lattice data are given for temperatures lower than 2.4 @xmath3 ; data are taken from @xcite.,width=321 ] the positive - definite function @xmath63 is only present as an overall factor in @xmath48 , so it does not come into play in the qualitative features of the effective potential .
however , it is relevant in view of reproducing the absolute value of the pressure in pure gauge qcd , for which lattice data are known at @xmath83 3 , 4 , 5 , 6 , 8 and @xmath84 through an extrapolation of these data @xcite .
the empirical choice @xmath85 leads to a good agreement between the lattice data of @xcite and formula ( [ pg ] ) , as shown in fig .
notice that the value @xmath86 is such that the stefan - boltzmann limit is reached at large temperatures .
pure gauge pressure computed from eq .
( [ pg ] ) and normalized to @xmath87 ( solid line ) .
the corresponding lattice data , taken from @xcite , are plotted for comparison in the case @xmath7 ( gray points ) and @xmath29 ( black points).,width=321 ] it is worth summarizing what has been done at this stage .
starting from lagrangian ( [ v1 ] ) , we have shown that the three arbitrary functions of @xmath2 and @xmath1 it contains can be strongly constrained by demanding that the averaged polyakov loop and the pure gauge pressure have a relevant behaviour in @xmath3 , at large @xmath2 , and in the large-@xmath1 limit .
explicit forms for the two remaining unconstrained functions of @xmath2 can then be found by asking the present model to be in agreement with current pure gauge lattice data .
one is finally left with a fully determined lagrangian with explicit z@xmath0 symmetry at finite @xmath1 and u(1 ) symmetry in the large-@xmath1 limit .
this lagrangian is obviously not predictive concerning the thermodynamics of the pure gauge sector , just as previously used lagrangians like ( [ ud ] ) .
however , its knowledge is a necessary step in view of making predictions concerning the quark sector , whose inclusion is discussed in the next section .
as shown in @xcite , a minimal coupling of the njl lagrangian ( [ njll ] ) to a gauge field of the form @xmath28 makes eventually appear the polyakov loop in the quark grand potential . in the mean field approximation ,
one is led indeed to the quark potential @xcite @xmath88+\frac{t}{n_c}{\rm tr}_c{\rm ln}\left [ 1+l^\dagger\ , { \rm e}^{-(e_p+\mu)/t}\right ] \right\rbrace \nonumber,\end{aligned}\ ] ] where the polyakov loop @xmath23 has been defined in ( [ l ] ) . in the above equality , @xmath89 is the quark dispersion relation , with @xmath26 the quark bare mass and @xmath90 related to the chiral condensate as follows @xmath91 the coupling @xmath92 has to scale as @xmath93 in order for the potential ( [ vq ] ) to scale as @xmath94 , so it is convenient to define the coupling @xmath10 as @xmath95 although @xmath96 is a priori arbitrary , our results are mostly valid at @xmath97 . for higher values of @xmath96 indeed , the axial anomaly ( not present in this formalism ) should be taken into account in order to get a reliable model . in what follows , @xmath97 will be implicitly understood , although we keep the notation @xmath96 so that the quark contributions appear more clearly .
since the pure gauge part of the potential only involves the traced polyakov loop @xmath22 , it is interesting to express @xmath98 in terms of @xmath22 rather than @xmath23 .
terms of the form @xmath99 $ ] can be expressed as functions of @xmath100 , @xmath101 , @xmath102 , through a taylor expansion .
a possible way of proceeding is to expand the quark potential at the first order in @xmath23 .
this eventually leads to formulas in which only @xmath22 appears in @xmath98 @xcite .
this scheme has the advantage of being independent of the parametrization of @xmath23 .
here we adopt an inequivalent procedure . as a first step ,
we notice that there exists in general a gauge in which the polyakov loop @xmath23 is a diagonal element of @xmath103 : @xmath104 the @xmath105 parameters @xmath106 are real so that @xmath107 and @xmath108 as demanded for an su(@xmath1 ) element . in the special case of @xmath7 ,
there is a one - to - one correspondence between the parameters @xmath109 , @xmath110 and the polyakov loop degrees of freedom @xmath22 , @xmath111 .
this is not the case at large @xmath1 however , where the number of independent parameters in the polyakov loop goes to infinity . as a consequence ,
an exact computation of the color traces appearing in ( [ vq ] ) is not possible unless simplifying assumptions are made . as a second step to reach this goal , we propose the following ansatz : @xmath112 it reduces to the mean - field parametrization of @xcite at @xmath7 , but the price to pay is that the number of degrees of freedom in @xmath23 is drastically reduced to a single real parameter @xmath113 .
it is then readily computed that @xmath114 & \qquad { \rm odd}-n_c \\ & = \cos\theta & \qquad { \rm even}-n_c \nonumber\end{aligned}\ ] ] by using of the ansatz ( [ ldef ] ) in ( [ phi ] ) .
we thus have an asatz that looks like " the su(3 ) case and that reduces to @xmath115 at large-@xmath1 .
moreover , one can compute that @xmath116&=&{\rm ln\ , det}_c\left [ 1+l\ , { \rm e}^{-(e_p-\mu)/t}\right]\nonumber\\ & = & \frac{n_c-1}{2}\ , { \rm ln}\left[1 + 2\frac{n_c\phi-1}{n_c-1}{\rm e}^{-(e_p-\mu)/t}+{\rm e}^{-2(e_p-\mu)/t } \right ] \nonumber\\ & & \ + { \rm ln}\left[1+{\rm e}^{-(e_p-\mu)/t}\right ] \qquad { \rm odd}-n_c , \\ & = & \frac{n_c}{2}\ , { \rm ln}\left[1 + 2\phi{\rm e}^{-(e_p-\mu)/t}+{\rm e}^{-2(e_p-\mu)/t } \right ] \nonumber\\ & & \qquad\qquad\qquad\qquad\qquad\quad\ , { \rm even}-n_c , \nonumber\end{aligned}\ ] ] and , taking into account a cutoff for the momentum integration of the vacuum term , one finally arrives at the quark potential , whose large-@xmath1 limit is given by @xmath117\right .
\nonumber\\ & & \left .
+ ( \mu\rightarrow -\mu ) \right\rbrace .\end{aligned}\ ] ] this last potential formally reduces to the genuine njl potential once @xmath118 , as observed in previous studies @xcite .
the total potential of the large-@xmath1 pnjl model under study is finally given by @xmath119 in the confined phase , where @xmath120 , one observes a term in @xmath121 $ ] in the potential ( [ vq2 ] ) , so it could be tempting to associate such a term with diquark degrees of freedom in the confined phase .
however , in the limit @xmath122 , one exactly recovers the zero temperature njl potential , expressed in terms of the quark degrees of freedom .
so the confined degrees of freedom are still quarks in the present approach .
similarly , baryonic degrees of freedom ( states with @xmath1 quarks ) are not included in the present formalism ; this would require further extensions of the present approach that are beyond the scope of the present study .
in t hooft s large-@xmath1 limit , the number of quark flavours stays finite and @xmath123 is dominated by the gluonic contribution .
consequently , when @xmath1 becomes infinite , the optimal value @xmath124 can be found by minimizing @xmath125 only . according to ( [ v30 ] )
, the large-@xmath1 solution ( _ i.e _ when @xmath1 is infinite ) reads @xmath126 the physical value of @xmath90 , denoted @xmath127 and depending on @xmath2 and @xmath4 , is then such that it minimizes @xmath128 .
@xmath125 does not depend on @xmath90 .
since @xmath129 , chiral symmetry is present when @xmath130 and broken when @xmath131 . as a consequence of ( [ phif ] ) , the deconfined phase appears as soon as @xmath80 , independently of the value of @xmath4 : as pointed out in @xcite , quarks have no influence on the deconfinement phase transition at large-@xmath1 because of the suppression of internal quark loops in this limit . as a consequence of the large-@xmath1 limit
, the confined / deconfined phases are straightforwardly identified in our model .
the situation is less simple as far as chiral symmetry is concerned ; numerical computations are needed . as a first step ,
the parameters of the model have to be fixed .
the values @xmath132 used in the pnjl study @xcite , will be taken in the following also .
the first three parameters have been fitted so that the zero - temperature pion mass and decay constant are reproduced within the standard njl model with @xmath7 and @xmath97 @xcite .
@xmath3 is a typical value for the deconfinement temperature in su(@xmath1 ) yang - mills theory . using the parameters ( [ p1 ] ) ,
the optimal value @xmath127 can now be computed for any couple @xmath133 , and can be linked to the quark condensate thanks to ( [ qq0 ] ) @xmath134 in the limit where @xmath2 and @xmath4 both tend toward zero , we get @xmath135 corresponding to a quite common value of @xmath136(317 mev)@xmath137 for @xmath7 and @xmath97 .
( @xmath29 ) , normalized to its zero temperature value , versus @xmath2 in units of @xmath3 , and plotted for @xmath138 0 , 1 , and 1.11 ( solid lines ) .
the optimal value of the polyakov loop is also plotted ( dotted line).,width=359 ] the large-@xmath1 chiral condensate versus the temperature is plotted in fig .
[ fig5 ] for some values of the quark chemical potential .
the most salient feature of this plot is the simultaneity of the first - order deconfinement phase transition and of the restoration of chiral symmetry through a first - order phase transition occurring at @xmath139 .
however , when @xmath140 ( @xmath141 200 mev ) , the quick decrease of the chiral condensate suggests a progressive restoration of chiral symmetry through a crossover at temperatures smaller than @xmath3 .
as shown in @xcite , the crossover temperature can be computed thanks to the determination of the peak position in the dimensionless quark susceptibility reading , at large-@xmath1 , @xmath142 we have chosen to follow the definition of @xcite for the quark susceptibility , although a more standard way of defining the susceptibility is rather @xmath143 , see _
e.g. _ @xcite . in both cases ,
a peak in the quark susceptibility signals a phase transition . a plot of @xmath144 for some values of @xmath145 is given in fig .
[ fig4b ] .
several observations can be made by observing this figure together with fig .
first , the peak of the quark susceptibility is located in @xmath3 when @xmath146 ; this corresponds to a first - order - type chiral symmetry restoration in the deconfined phase .
the point @xmath147 actually corresponds to a triple point in the @xmath133-plane : at large @xmath4 the peak of @xmath148 is located below @xmath3 a larger @xmath4 corresponds to a lower peak position , leading to the existence of a confined phase in which chiral symmetry is progressively restored through a crossover
. a careful look at @xmath127 actually shows that the chiral phase transition below @xmath3 becomes of first order when @xmath149 : there exists thus a critical - end - point that we find to be @xmath150 in the @xmath133-plane .
apart from the susceptibility , the position of the chiral phase transition could have been alternatively determined by computing the zero of @xmath151 .
we have checked that the chiral temperatures computed using that method agree with those computed with those computed from the peak in the susceptibility up to 5% . for an exploratory study such as the present one ,
this agreement is satisfactory . ) versus the temperature in units of @xmath3 ( solid lines ) with , from left to right , @xmath1381.24 , 1.15 , 1.04 , 0.93 , 0.79 .
@xmath152 is also plotted for completeness ( dashed line ) .
computations were done for @xmath29 .
, width=321 ] gathering all these observations , the phase diagram of our model in the @xmath133-plane can be established ; it is shown in fig .
the three phases we find correspond to those found in @xcite , see fig .
[ fig1 ] , but the structure of the chiral phase transition is a bit more involved under @xmath3 : the chemical potential at which chiral symmetry is restored now depends on @xmath2 , and there exists a critical - end - point at large enough @xmath4 .
although the deconfining phase transition corresponds to what is expected in the large-@xmath1 limit of qcd from generic arguments @xcite , the critical line @xmath6 we find under @xmath3 quite resembles to what can be observed within previously known @xmath7 pnjl studies @xcite .
the similarity between our way of including the polyakov loop in the njl model and the way of @xcite our ansatz is a straightforward generalization of the one used in this last work
might actually be at the origin of the similarities between the phase diagrams we find .
the same reason , combined to the fact that we chose for our parameters values fitted on the su(3 ) case , might explain why the values we find for _ e.g. _ the chiral condensate are similar to those of @xcite .
pnjl model ( [ potf ] ) with explicit z@xmath0 symmetry , obtained for @xmath29 .
the solid lines denote first - order phase transitions while the dashed line denotes a crossover . the triple point ( 0.212 , 0.270 )
gev and the critical end - point ( 0.335 , 0.063 ) gev have been also plotted .
the end of the lower curve is reached at ( 0.343 , 0 ) gev.,width=321 ] we notice that , at large but finite values of @xmath1 , the full potential ( [ potf ] ) has to be minimized and quark contributions ( presumably in @xmath153 ) will cause the polyakov loop to be different from @xmath64 . hence , the chiral condensate will also be modified , and the whole phase diagram will be affected .
we nevertheless choose here to focus on the large-@xmath1 limit of the model , since it has been designed to be relevant in this limit mostly .
effective polyakov - loop - based " approaches have proven to be a relevant tool in view of modelling the thermodynamic properties of pure gauge qcd .
the traced polyakov loop is then the order parameter associated to confinement , itself seen as correlated with a global center symmetry , z@xmath0 when the gauge group is su(@xmath1 ) . following a suggestion made in @xcite , an explicitly z@xmath0-symmetric potential involving the traced polyakov loop has been built and leads to : a first - order phase transition at large-@xmath1 , a gluonic pressure scaling as @xmath154 , and an @xmath1-independent optimal value for the polyakov loop .
the coupling of the pure gauge sector to light quarks has then been performed within a pnjl approach .
thanks to a particular ansatz for the polyakov loop , the quark potential is such that it only involves the traced polyakov loop that appears in the pure gauge potential .
it has to be said that the resulting pnjl model is designed to be relevant in the large-@xmath1 limit only .
it is moreover inequivalent to the model proposed in @xcite since the colour traces that have to be performed in the derivation of the pnjl potential are carried differently .
the different assumptions that were made lead to a large-@xmath1 phase diagram that compares favourably with previous approaches . at any @xmath4
, it shows a deconfined , chirally symmetric , phase above the deconfinement temperature ( @xmath155270 mev ) .
the deconfinement phase transition and the restoration of chiral symmetry are found to be simultaneous first - order phase transitions , in agreement with @xcite . at temperatures lower than @xmath3 ,
thus in the confined phase , there is a critical line @xmath6 separating a phase with broken chiral symmetry at small @xmath4 and a chirally symmetric phase at large @xmath4 .
the phase transition is found to be a crossover from the triple point ( 0.212 , 0.270 ) gev to the critical - end - point ( 0.335 , 0.063 ) gev .
it is then of first order until the boundary ( 0.343 , 0 ) gev is reached , corresponding to the estimate @xmath156 @xcite .
it is worth saying that a confined , chirally symmetric , phase has also been found by solving schwinger - dyson equations at nonzero @xmath4 in @xcite , and that evidences for the existence of such a phase has been found in coulomb gauge qcd calculations @xcite .
the present model s prediction of three different phases , among which a chirally symmetric phase at baryonic chemical potentials around the nucleon mass thus seems a reliable one .
the structure of the critical lines appears however to depend quite strongly on the way the polyakov loop @xmath23 is handled : the comparison of figs .
[ fig1 ] and [ fig6 ] shows how a different treatment of the polyakov loop within similar pnjl - based approaches can affect the phase diagram .
it is also worth noting that an accurate description of the confined but chirally symmetric
quarkyonic at large @xmath1 phase would require a more detailed approach that the simple pnjl model presented here .
in particular , the explicit inclusion of baryonic degrees of freedom should be performed , but this is a task that we leave for subsequent studies .
a first interesting attempt has for example been made in @xcite , where a constituent approach is used to model the baryonic matter at large @xmath1 .
recently , an effective model has been proposed @xcite , in which the lagrangian involves a linear sigma model for the quark part and a dilaton - like effective potential for the gluon part . such a dilaton - like potential is designed to mimic the breaking of scale invariance through a nonzero value of the gluon condensate ( the dilaton field ) .
the phase diagram which is found is nearly identical to the one of fig .
[ fig6 ] , excepted that the deconfinement phase transition is replaced by a restoration of scale invariance .
both seem then to be linked , as suggested in @xcite .
finally , we remark that the large-@xmath1 limit of the proposed pure gauge effective potential has a u(1 ) symmetry , which emerges as the limit of a z@xmath0 symmetry .
this large-@xmath1 effective potential could be used in an approach where the traced polyakov loop is allowed to depend on the position , typically via a lagrangian of the type @xmath157 .
of particular interest would then be to search for localised , solitonic , solutions of @xmath158 : one could then take advantage of the fact that finding solutions of a complex scalar field theory with a u(1 ) invariance is a topic that has attracted a considerable attention , mostly since coleman s work on q - balls @xcite , and for which many results are already available .
we hope to present such a study in future works .
99 m. a. stephanov , pos * lat2006 * , 024 ( 2006 ) [ hep - lat/0701002 ] ; k. fukushima and t. hatsuda , rept .
* 74 * , 014001 ( 2011 ) [ arxiv:1005.4814 ] .
l. susskind , phys .
d * 20 * , 2610 ( 1979 ) ; n. weiss , phys .
d * 24 * , 476 ( 1981 ) ; n. weiss , phys .
d * 25 * , 2668 ( 1982 ) .
b. svetitsky and l. g. yaffe , nucl .
b * 210 * , 423 ( 1982 ) ; l. g. yaffe and b. svetitsky , phys .
d * 26 * , 963 ( 1982 ) .
r. d. pisarski , phys .
d * 62 * , 111501 ( 2000 ) [ hep - ph/0006205 ] .
a. dumitru and r. d. pisarski , phys
. lett .
b * 504 * , 282 ( 2001 ) [ hep - ph/0010083 ] ; e. s. fraga , r. d. pisarski and j. schaffner - bielich , phys .
d * 63 * , 121702 ( 2001 ) [ hep - ph/0101143 ] ; o. scavenius , a. dumitru and j. t. lenaghan , phys .
c * 66 * , 034903 ( 2002 ) [ hep - ph/0201079 ] .
y. nambu , g. jona - lasinio , phys .
* 122 * , 345 ( 1961 ) ; y. nambu , g. jona - lasinio , phys . rev . *
124 * , 246 ( 1961 ) .
s.p . klevansky , rev .
* 64 * , 649 ( 1992 ) [ hep - ph/9401310 ] .
k. fukushima , phys .
b * 591 * , 277 ( 2004 ) [ hep - ph/0310121 ] . c. ratti , m. a. thaler and w. weise , phys .
d * 73 * , 014019 ( 2006 ) [ hep - ph/0506234 ] .
a. dumitru , d. roder and j. ruppert , phys .
d*70 * , 074001 ( 2004 ) [ hep - ph/0311119 ] ; s. k. ghosh , t. k. mukherjee , m. g. mustafa and r. ray , phys .
d*73 * , 114007 ( 2006 ) [ hep - ph/0603050 ] ; c. sasaki , b. friman and k. redlich , phys . rev .
d*75 * , 074013 ( 2007 ) [ hep - ph/0611147 ] ; b. -j .
schaefer , j. m. pawlowski and j. wambach , phys . rev .
d*76 * , 074023 ( 2007 ) [ arxiv:0704.3234 ] ; w. -j .
fu , z. zhang and y. -x .
liu , phys .
d * 77 * , 014006 ( 2008 ) [ arxiv:0711.0154 ] ; k. fukushima , phys .
d * 77 * , 114028 ( 2008 ) [ arxiv:0803.3318 ] ; s. roessner , t. hell , c. ratti and w. weise , nucl .
a * 814 * , 118 ( 2008 ) [ arxiv:0712.3152 ] ; h. nishimura and m. c. ogilvie , phys .
d * 81 * , 014018 ( 2010 ) [ arxiv:0911.2696 ] ; k. fukushima , m. ruggieri and r. gatto , phys . rev .
d * 81 * , 114031 ( 2010 ) [ arxiv:1003.0047 ] ; a. j. mizher , m. n. chernodub and e. s. fraga , phys .
d * 82 * , 105016 ( 2010 ) [ arxiv:1004.2712 ] ; r. gatto and m. ruggieri , phys .
d * 82 * , 054027 ( 2010 ) [ arxiv:1007.0790 ] .
d. blaschke , m. buballa , a. e. radzhabov , m. k. volkov , yad .
fiz . * 71 * , 2012 ( 2008 ) [ arxiv:0705.0384 ] ; g. a. contrera , d. gomez dumm and n. n. scoccola , phys .
b * 661 * , 113 ( 2008 ) [ arxiv:0711.0139 ] ; t. hell , s. roessner , m. cristoforetti , w. weise , phys .
d * 79 * , 014022 ( 2009 ) [ arxiv:0810.1099 ] ; m. buballa , a. g. grunfeld , a. e. radzhabov , d. scheffler , prog .
phys . * 62 * , 365 - 370 ( 2009 ) [ arxiv:0811.4543 ] ; t. hell , s. rossner , m. cristoforetti , w. weise , phys .
d * 81 * , 074034 ( 2010 ) [ arxiv:0911.3510 ] ; a. e. radzhabov , d. blaschke , m. buballa and m. k. volkov , phys .
d * 83 * , 116004 ( 2011 ) [ arxiv:1012.0664 ] ; d. horvatic , d. blaschke , d. klabucar , o. kaczmarek , phys .
d * 84 * , 016005 ( 2011 ) [ arxiv:1012.2113 ] ; t. hell , k. kashiwa , w. weise , phys .
d * 83 * , 114008 ( 2011 ) .
[ arxiv:1104.0572 ] .
g. a. contrera , m. orsaria and n. n. scoccola , phys .
d * 82 * , 054026 ( 2010 ) [ arxiv:1006.4639 ] .
l. mclerran , k. redlich and c. sasaki , nucl .
a*824 * , 86 ( 2009 ) [ arxiv:0812.3585 ] .
l. mclerran and r. d. pisarski , nucl .
a*796 * , 83 ( 2007 ) [ arxiv:0706.2191 ] .
h. abuki , r. anglani , r. gatto , g. nardulli and m. ruggieri , phys .
d*78 * , 034034 ( 2008 ) [ arxiv:0805.1509 ] ; a. andronic _ et al . _ , nucl .
a*837 * , 65 ( 2010 ) [ arxiv:0911.4806 ] .
f. sannino , phys .
d * 72 * , 125006 ( 2005 ) [ hep - th/0507251 ] .
as pointed out in @xcite , the term @xmath159 possesses the same symmetry but violates charge conjugation , which is also a symmetry of the effective theory under consideration .
s. roessner , c. ratti , w. weise , phys .
d*75 * , 034007 ( 2007 ) [ hep - ph/0609281 ] .
j. polnyi and k. szlachnyi , phys .
b * 110 * , 395 ( 1982 ) ; m. gross , phys .
b * 132 * , 125 ( 1983 ) .
y. makeenko , _ methods of contemporary gauge theory _ ( cambridge university press , 2002 ) .
b. lucini , m. teper and u. wenger , jhep * 0502 * , 033 ( 2005 ) [ hep - lat/0502003 ] .
j. braun , a. eichhorn , h. gies , j. m. pawlowski , eur .
j. c*70 * , 689 ( 2010 ) [ arxiv:1007.2619 ] .
s. gupta , k. huebner and o. kaczmarek , phys .
d * 77 * , 034503 ( 2008 ) [ arxiv:0711.2251 ] .
m. panero , phys .
* 103 * , 232001 ( 2009 ) [ arxiv:0907.3719 ] .
t. hatsuda and t. kunihiro , phys . rept . *
247 * , 221 ( 1994 ) [ hep - th/9212088 ] .
l. y. glozman and r. f. wagenbrunn , phys .
d * 77 * , 054027 ( 2008 ) [ arxiv:0709.3080 ] .
p. m. lo and e. s. swanson , phys .
d * 81 * , 034030 ( 2010 ) [ arxiv:0908.4099 ] .
t. d. cohen , n. kumar and k. k. ndousse , phys .
c * 84 * , 015204 ( 2011 ) [ arxiv:1102.2197 ] . c. sasaki and i. mishustin , arxiv:1110.3498 .
s. coleman , nucl .
b * 262 * , 263 ( 1985 ) .
|
a pnjl model is built , in which the polyakov - loop potential is explicitly z@xmath0-symmetric in order to mimic a yang - mills theory with gauge group su(@xmath1 ) .
the physically expected large-@xmath1 and large-@xmath2 behaviours of the thermodynamic observables computed from the polyakov - loop potential are used to constrain its free parameters .
the effective potential is eventually u(1)-symmetric when @xmath1 is infinite .
light quark flavours are added by using a nambu - jona - lasinio ( njl ) model coupled to the polyakov loop ( the pnjl model ) , and the different phases of the resulting pnjl model are discussed in t hooft s large-@xmath1 limit .
three phases are found , in agreement with previous studies resorting to effective approaches of qcd .
when the temperature @xmath2 is larger than some deconfinement temperature @xmath3 , the system is in a deconfined , chirally symmetric , phase for any quark chemical potential @xmath4 . when @xmath5 however , the system is in a confined phase in which chiral symmetry is either broken or not . the critical line @xmath6 , signalling the restoration of chiral symmetry , has the same qualitative features than what can be obtained within a standard @xmath7 pnjl model .
|
various mesomorphic structures can be prepared using the strong ( ionic ) or weak ( hydrogen ) bonding , respectively , between homopolymers and head functionalized oligomers or surfactants @xcite .
the interaction between the head group of the oligomeric molecule and suitable groups on the homopolymer chain , on the one hand , and unfavorable polar nonpolar interactions between the non polar tail of the oligomer and the rest of the system , on the other one , can cause the microphase separation .
the latter effect results in a periodic alternation of the layers of stretched homopolymer chains surrounded by perpendicularly oriented oligomeric tails as it is shown in figure 1 .
an example of the ionically bonded system is represented by the homopolymer - oligomer mixture of atactic poly(4-vinyl pyridine ) ( p4vp ) and dodecyl benzene sulfonic acid ( dbsa ) where due to a very strong interaction the microphase separation is realized over the whole temperature region @xcite .
the peculiarity of the systems of this kind is an increasing long space period @xmath0 as a function of the oligomeric fraction @xmath1 defined as the number of the oligomeric ( dbsa ) molecules per one pyridine ring ( see figure 2 ) .
more complicated behavior is inherent in the hydrogen bonded systems where the weak interaction causes an order disorder transition to homogeneous high temperature state @xcite .
an example of the system of this kind represents the mixture p4vp(pdp)@xmath2 of the same homopolymer p4vp with the surfactant 3-pentadecyl phenol ( pdp ) . here
, contrary to the ionically bonded systems , the long space period decreases with increasing @xmath1 ( see figure 3 ) . as it is shown in figure 2 , an intermediate behavior ( non monotonous @xmath1dependence of the long space period ) was found in the system p4vp-(zn(dbs)@xmath3)@xmath2 where the ionic interaction between zinc dodecyl benzene sulfonate zn(dbs)@xmath3 and p4vp is somewhat weaker than in the p4vp-(dbsa)@xmath2 system@xcite . principally important for our considerations
is the decreasing character of the temperature dependence of the long period found experimentally for all the systems mentioned above @xcite @xcite .
however , such a character of the dependence appears in the hydrogen bonded systems only within a finite temperature interval limited by the temperatures of crystallization ( or glass transition @xmath4 ) from below and of the order disorder transition @xmath5 from above ( see figure 4 ) .
as it is shown in figure 5 , an increase of the oligomeric fraction leads to a non monotonous dependence of the temperature @xmath5 with a maximum occuring near @xmath6 .
the aim of this paper is to explain peculiarities of the microphased separated homopolymer
oligomer mixtures with both , strong and weak bonding , within a framework of the unified scheme of stochastic systems @xcite .
our consideration is based on the obvious equality for the long period @xmath7 where @xmath8 is the oligomeric chain length and @xmath9 is the thickness of the homopolymer layer being fixed due to sharing of the surfactant molecules by homopolymer chains ( see figure 1 ) .
physically , this share is reduced to the inverse magnitude @xmath10 of the circular frequency of the alternation of the oligomer heads along the homopolymer chain .
then , the long period can be expressed by the following equation @xcite @xmath11 where @xmath12 is the flory interaction parameter ( @xmath13 is the degree of polymerization of the homopolymer chains ) , @xmath14 is the number of segments in the oligomer chain , @xmath15 is the segment length .
the frequency @xmath16 will be found from the effective equation of motion which is stochastic in nature and where the segment number @xmath17 ( @xmath18 ) plays the role of imagined time @xcite .
the problem under consideration is addressed by the definition of effective law of motion @xmath19 that determines the sequence of the alternation of oligomers along the homopolymer chain by means of the occupation number , @xmath19 , being @xmath20 if the oligomer is attached to the segment @xmath17 , and @xmath21 otherwise . in the limit @xmath22 , the argument @xmath17 may be considered as a continuous one , and the behavior of the system is governed by the action and the dissipative functional @xmath23 with dimensionless lagrangian @xmath24 here dot denotes the derivative with respect to the segment number @xmath17 , @xmath25 is temperature in energy units , @xmath26 , @xmath27 and @xmath28 are inverse kinetic coefficient , inhomogeneity and interaction parameters , respectively . the key point in our consideration
consists in that the effective mass @xmath27 in eq.([4 ] ) is a fluctuating parameter with the mean value @xmath29 and the variance @xmath30 ( bar denotes the average , as usually ) .
then , after the averaging of the exponent @xmath31 over gaussian distribution of the bare mass @xmath27 , we obtain renormalized inhomogeneity energy @xmath32 as a result , taking into account the stochastic source @xmath33 , relevant euler equation arrives at non linear langevin equation @xmath34 where the effective mass @xmath35 , characteristic number of correlating segments @xmath36 and @xmath37correlated noise @xmath38 , respectively , are introduced in accordance with definitions : @xmath39 to linearize the equation of motion ( [ 8 ] ) within the self consistent approach , it is necessary to replace the fluctuational term @xmath40 of the renormalized mass in eq .
( [ 7 ] ) by the averaged expression @xmath41 where @xmath42 is the structure factor in the frequency representation ( angle brackets denote average over noise @xmath33 ) .
then , the green function @xmath43 and the structure factor @xmath44 take the forms : @xmath45^{-1 } , \quad s=\left[\tau^2(\nu)+n_c^{2}\nu^2\right]^{-1 } \label{12}\ ] ] where renormalized interaction parameter is introduced by * equalities * @xmath46
to obtain physically observable values , at first one has to determine the effective mass @xmath47 given by eqs.([13 ] ) , ( [ 11 ] ) . on the basis of the theory of residues , using the structure factor , eq.([12 ] ) arrives at the expression for renormalization mass parameter @xmath48 inserting here eqs.([9 ] ) , ( [ 13 ] ) , we obtain the effective inhomogeneity parameter @xmath49 where the temperature domain is bounded above by the characteristic temperature @xmath50 according to eq.([15 ] ) the effective mass decreases monotonously with increasing temperature from the bare magnitude @xmath51 at @xmath52 to @xmath53 at @xmath54 .
the divergency condition of the green function ( [ 12 ] ) arrives at the proper frequency of the oligomer alternation @xmath55 with imaginary and real parts @xmath56 insertion of eqs.([9 ] ) , ( [ 15 ] ) gives the temperature dependence of the proper frequency @xmath57 where the bare frequency @xmath58 and another characteristic temperature @xmath59 are introduced as follows : @xmath60 as a result , combination of eqs.([1 ] ) , ( [ 19 ] ) leads to the final result @xmath61 where the characteristic length , @xmath62 , is defined as @xmath63
the behavior of the system is determined by the relation @xmath64 where the minimal magnitude @xmath65 fixes the choice of the theory parameters according to the condition : @xmath66 .
it would seem from eq.([23 ] ) that the decrease of the temperaure @xmath67 with weakening interaction when passing from the ionically bonded system ( such as p4vp-(dbsa)@xmath2 ) to the hydrogen bonded one ( e.g. , p4vp(pdp)@xmath2 ) is caused only by the growth of the fluctuation parameter @xmath68 with respect to the mean magnitude of the inhomogeneity parameter @xmath29 .
it appeares , however , that the main reason for such behavior is a decrease of the mean geometrical magnitude @xmath69 of the principle coefficients in the generic lagrangian ( [ 4 ] ) . to clarify this problem ,
assume the three
parametric @xmath1dependencies for the above parameters @xmath70 with positive constants @xmath71 , @xmath72 , @xmath73 , @xmath74 , @xmath75 , @xmath76 . in limiting case , @xmath77 relevant to the ionically bonded systems , when @xmath78 , @xmath79 , the temperature @xmath5 is so large that the temperature dependence ( [ 21 ] ) takes the form @xmath80 then , fitting of the experimental data shown in figure 2 according to eq.([26 ] ) arrives at the following results .
the system p4vp-(dbsa)@xmath2 : @xmath81 , @xmath82 , @xmath83 ;
@xmath84 , @xmath85 , @xmath86 ; @xmath87 nm ; @xmath88 nm .
the system p4vp-(zn(dbs)@xmath3)@xmath2 : @xmath89 , @xmath90 , @xmath91 ;
@xmath84 , @xmath92 , @xmath93 ; @xmath87 nm ; @xmath88 nm . respectively , for @xmath94 eq.([23 ] ) gives values @xmath95 at ratios @xmath96 .
a much more complicated situation occurs in hydrogen bonded system p4vp(pdp)@xmath2 , where the reduction of the parameter ( [ 23 ] ) leads to a more narrow temperature domain @xmath97 . as a result , the temperature dependence of the period @xmath98 obtains a more complicated form ( [ 21 ] ) which keeps the decreasing character as shown in figures 3 and 4 .
the dependence @xmath99 at a fixed temperature may be described by the estimation @xmath100 following from eq.([21 ] ) and conditions @xmath101 , @xmath102 .
then , the experimental data shown in figure 3 give the following constrains : @xmath103 , @xmath104 , @xmath105 . respectively , the comparison of the experimental data shown in figure 4 with the fitting results using eqs.([15 ] ) , ( [ 16 ] ) , ( [ 20 ] ) and ( [ 21 ] ) yields @xmath106 , @xmath107 nm .
finally , figure 5 and eq.([16 ] ) give @xmath108 , @xmath109 , @xmath110 . as a result ,
taking @xmath111 at @xmath94 the magnitudes @xmath112 , @xmath113 , @xmath114 are obtained and they provide the value of @xmath72 smaller than @xmath115 which arrive at the parameter @xmath116 and ratio @xmath117 .
it is worth to mention that the value of the parameter @xmath118 is extremally small that guarantees the validity of the flory
huggins approach .
|
microphase separated structure consisting of the periodic alternation of the layers of stretched homopolymer chains surrounded by perpendicularly oriented oligomeric ( surfactant ) tails is studied for the systems with strong ionic ) as well as weak ( hydrogen ) interaction .
our approach is based on the fact that the structure period is determined by alternating associations between the head groups of the oligomer molecules and interacting groups of the homopolymer chains .
distribution of oligomers along the homopolymer chains is described by the effective equation of motion with the segment number playing the role of time . as a result ,
experimentally observed temperature dependence of the structure period , as well as the dependence of the point of order disorder transition are determined as functions of the oligomeric fraction .
|
collective spin dynamics in dilute magnetic semiconductors ( dms ) is an active field under intense investigations.@xcite this field provides an insight into the origins of carrier - induced ferromagnetism in semiconductors @xcite and an understanding of particular features of the dms @xcite due to the presence of two spin sub - systems that are dynamically coupled via coulomb - exchange interaction : that of the itinerant carrier and that of the localized magnetic impurities . as an example of these features , the transverse spin excitation spectrum has been theoretically found to be composed of three types of excitations .
these are : two collective excitations corresponding to itinerant and localized spins precessing in phase or out of phase to each other , and single - particle ( or stoner - like ) excitations of the itinerant carriers.@xcite the in - phase collective mode is the goldstone - like mode .
the out of phase mode has a dominant contribution from the itinerant carrier subsystem.@xcite experimental evidences of the entire spectrum in ferromagnetic dms like gamnas is not available .
reported so far are features related to the zone - center in phase mode , dominated by the mn spin precession , its dynamics@xcite and its ferromagnetic resonance.@xcite we find no experimental data available for the out of phase mode .
indeed , ferromagnetism in gamnas systems requires a high mn concentration , which destroys the quality of the crystal potential and smooths out all optical resonances . to gain more insight into the dms spin excitation spectrum , cd@xmath0mn@xmath1te doped quantum wells are a very good alternative as they are clean and efficient to capture the general properties of the collective spin dynamics in dms materials .
the high quality of cd@xmath0mn@xmath1te quantum well structures leads to a high mobility of the itinerant gas ( electrons or holes ) together with well - defined optical resonances,@xcite an important condition for the observation of electronic raman signals from the carrier subsystem .
@xcite the presence of mn impurities interacting with the two dimensional electron gas ( 2deg ) embedded in the cd@xmath0mn@xmath1te quantum well , thus , provides interest on two levels .
first , the static mean - field _ s - d _ contribution from the mn ions produces a large zeeman splitting : @xmath2 where @xmath3 is the overhauser shift due to the mn spin polarization and @xmath4 is the normal zeeman splitting of the conduction band ( here @xmath5 for @xmath6 ) .
this leads to a high spin polarization degree @xmath7 in the 2deg.@xcite it makes it a suitable system to study features of spin - resolved coulomb interaction.@xcite the above approach was followed in our previous work to study spin excitations of the spin - polarized 2deg ( sp2deg).@xcite .
propagative spin flip waves ( sfw ) of electrons were observed @xcite and their appearance was linked to the coulomb interaction between electrons.@xcite on a second level , it provides evidence of the mixing between electrons and mn spin modes in the frequency@xcite and time@xcite domains as predicted in ref.@xcite .
nevertheless , a connection between these manifestations needs to be made : in the first approach , _ s - d _ dynamical coupling with the mn was neglected , while in the second , coulomb interaction between electrons did not manifest experimentally and was not taken into account in theories of ref.@xcite .
there is , therefore , a lack of full understanding of the mixed spin - excitations where both coulomb interactions between electrons and the _ s - d _ dynamical coupling with mn spins interplay .
moreover , in the later studies,@xcite only the zone center collective electron - mn modes were probed . from the larmor s
theorem,@xcite it is known that the spin system behaves as if the electrons were not interacting through coulomb .
this explains why theories describing the electron - mn modes , without the coulomb interaction between carriers,@xcite agree well with the experimental data of ref.@xcite . out of zone center
, predictions of ref.@xcite , found the out of phase mode ( so - called stiff mode ) to have a positive dispersion above the stoner continuum .
however , we have shown by angle - resolved raman measurements that the collective electron spin mode ( sfw mode ) , was below the spin - flip single - particle excitations ( sf - spe)@xcite , together with a negative slope dispersion with the raman transferred in plane momentum @xmath8.@xcite it is natural to make the correspondence between the stiff mode of ref.@xcite and the sfw mode of refs.@xcite and @xcite ( stoner - like excitations are obviously the sf - spe )
. therefore , our experimental observations are in contradiction with the theory of ref.@xcite . in the following
we will use the denomination `` out of phase mode '' or opm to describe a mode which coincides with the stiff mode when the coulomb interaction between carriers is neglected or to the sfw mode when the _ s - d _ dynamical coupling with mn spins is neglected . in this work ,
we want to resolve this contradiction , by first showing new experimental data that clearly shows the sf - spe continuum and that the opm propagates below it with a negative dispersion .
secondly , we will introduce a new propagator for the electron - mn collective modes that includes the coulomb interaction between spin - polarized carriers .
finally , we justify the experimental features by comparing the _ s - d _ exchange dynamical correction due to the coupling of the electron and mn spin precession with the correction introduced by the coulomb interaction between carriers .
we will demonstrate that the ratio of these two corrections does not depend , at standard conditions , on the mn concentration and that the coulomb correction always dominates over the _ s - d _ dynamical shift except for very large electron densities .
our measurements were made in the back scattering geometry on a cd@xmath0mn@xmath9te single quantum well with electron densities @xmath10 ranging from @xmath11 @xmath12 to @xmath13 @xmath12 and an effective mn@xmath14 concentration @xmath15 ranging from @xmath16 to @xmath17.@xcite the quantum wells were grown by molecular beam epitaxy on a gaas substrate.@xcite these quantum wells and their spin polarizations have been characterized in our previous publication.@xcite here we will consider samples a and b of that reference .
both a and b have a density of @xmath18 @xmath12 and different mn concentrations , @xmath19 and @xmath20 , respectively .
measurements were taken with the samples immersed in superfluid helium at @xmath21 k inside a superconducting magnetic cryostat .
the essential part of the cryostat is in the rod holding the sample , which allows for control in three directions , including rotation of the sample in the plane of the magnetic field for dispersions while the collection / reflection mirror was kept fixed .
this design aids in an easy study of the raman scattering dispersions with an accuracy of 1 degree ( @xmath22 @xmath23 ) . to determine the in - plane wave vector @xmath8 transferred in the plane of the quantum
well , the angle of rotation @xmath24 is measured between the normal of incidence of the sample and the scattered light .
@xmath8 is obtained from the geometry of the set - up as @xmath25 , where @xmath26 is the excitation wavelength.@xcite the holder was inserted in the core of a solenoid producing magnetic fields of up to 5 t in the plane of the quantum well .
a tunable cw ti - sapphire laser , pumped by an ar@xmath27 laser was used as light source .
the laser power density was kept below 0.1 w/@xmath28 and the linewidth was 0.3 @xmath29 at a slit opening of 50 @xmath30 m .
the energy of the laser was tuned to resonate close to the transition between the first conduction band and the second heavy hole band of the quantum well .
t , ( b ) @xmath31 t and ( c ) @xmath32 t in sample a. arrows indicate locations of sf - spe , spe1 and spe2 as described in the text .
the lines at 0 mev are remnants of the laser line . ] in fig .
[ fig : spectra ] , we show typical raman scattering spectra obtained on sample a for various raman transferred momentum @xmath8 obtained at different magnetic fields : ( a ) @xmath33 t , ( b ) @xmath31 t and ( c ) @xmath32 t. in the figure , @xmath8 varies from 0 to @xmath3415 @xmath30 m @xmath35 .
all spectra are dominated by a sharp - peaked line associated with the collective sfw as already demonstrated in ref.@xcite .
we will show in the following that this mode is exactly the out of phase electron - mn mode ( the stiff mode of ref.@xcite ) of the dms system .
the sfw mode is narrow in width compared to the other excitations in the spectra and is found to be intense at all magnetic fields .
as @xmath8 is increased , the intensity and energy of the sfw decreases and eventually diminishes at a critical wave vector , @xmath36.@xcite the linewidth of the sfw broadens with increasing @xmath8 , damping out due to interactions with the sf - spe continuum.@xcite at high energies , features of the sf - spe are observed in the spectra of fig .
[ fig : spectra ] .
these features emerge as the magnetic field is increased , showing a fan - shaped like behavior at large @xmath8 values leading to a continuum .
this continuum corresponds to the joint density of state of electron - hole pairs in the fermi sea which has a characteristic double - peak structure.@xcite in the following , we have labeled the maxima positions of the sf - spe continuum as spe1 and spe2 . in the long wavelength limit ,
spe1 and spe2 peaks are close to @xmath37 where @xmath38 is the fermi velocity of the minority spin subband and @xmath39 is the renormalized zeeman energy.@xcite spe1 and spe2 are degenerate at @xmath40 forming a single peaked structure positioned at @xmath39.@xcite at small wavevectors , a clear distinction between spe1 and spe2 is difficult as these features overlap .
spe1 shifts to low energies , decreasing in intensity and eventually vanishing as @xmath8 is increased .
spe2 on the other hand , increases both in intensity and energy when increasing @xmath8 .
the dip between spe1 and spe2 is due to the occupancy of the upper spin subband .
values taken on sample b with the corresponding calculated spectra ( b ) and ( c ) .
( b ) is obtained from the collective response and ( c ) from the single particle response . ] in fig .
[ fig : sampb&theory ] , we compare the raman scattering spectra taken on sample b ( @xmath41 and @xmath18 @xmath12 ) for various @xmath8 with calculated spectra for the same conditions obtained from the collective response and the single particle response of ref.@xcite .
the experimental spectra are neither given by the collective response , nor the single - particle one , but by a mixing of both responses .
this is a consequence of the optical resonance with the incoming raman photon.@xcite ( a ) and ( c ) .
these , respectively , correspond to spin polarization values @xmath42 and @xmath43 .
experimental data are compared with calculations : the dashed domain is the sf - spe continuum , edges of the domain are given for @xmath44 k. the lines inside the domain ( when available ) are the peaks of the single particle response [ see fig .
[ fig : sampb&theory](c ) ] calculated at @xmath45 k. the line below the continuum is the opm dispersion . ]
we now plot in fig .
[ fig : disper ] the dispersions of the sfw and features of the sf - spe extracted from fig .
[ fig : spectra ] and compare with calculations .
the spin polarizations in fig.[fig : disper](a ) and [ fig : disper](b ) are respectively @xmath46 and @xmath47 . in the spectra of fig .
[ fig : spectra ] and the dispersions in fig .
[ fig : disper ] , our data show clearly that the sfw mode propagates below the sf - spe continuum with a negative slope dispersion at energies that are in exact agreement with the calculated opm energy .
evidence is given that this experimental mode is the opm ( or stiff mode of ref.@xcite ) .
consequently , our observations are in contradiction to this theory which predicted that the dynamical _ s - d _ exchange interaction between the mn ions and the electrons shifts the opm above the sf - spe continuum with a positive dispersion . to resolve this contradiction ,
we have introduced the coulomb interaction between carriers to calculate the coupled electron - mn response .
details of the calculations are given in ref.@xcite .
the mixed response function is calculated adiabatically and leads to the following electron spin susceptibility : @xmath48 and wavevector @xmath8 . in the plane
@xmath40 , the two lower curves are the solutions @xmath49 of eq.([zonecentermodes ] ) .
the opm is the higher energy solution .
the upper curve is the sf - spe ( stoner - like excitations ) degenerate to @xmath39 . out of the resonant field ( @xmath50 ) where the modes anti - cross , the branches have electron or mn character . at the resonant field ( indicated by a vertical arrow ) , the avoided gap is about @xmath51ev .
in addition , dispersions @xmath52 obtained from eq.([eqcoulomb ] ) have been plotted for @xmath53 t and @xmath54 t. for @xmath53 t the electron or mn characters of the mode do not change . however , for @xmath54 t the modes anticross and an avoided gap appears at a finite @xmath55 ( indicated by the vertical arrow ) .
modes exchange their slope dispersions and characters when @xmath8 varies accross @xmath55 .
calculations were done for @xmath56 and a square quantum well of width @xmath57 nm . ] where @xmath58is the linear response of the transverse electron spin fluctuation observable @xmath59 to the perturbation @xmath60
created by a transverse magnetic field of temporal and spatial fourier components , respectively , given by @xmath61 and @xmath62 @xcite coupled electron - mn modes appear as poles of @xmath63 . in eq.([eqcoulomb ] ) @xmath64 is the mn spin precession energy in the presence of the 2deg equilibrium spin polarization , @xmath65 is the manganese @xmath66 factor ; @xmath67 is the bohr magneton ; @xmath68 is the electron effective mass ; @xmath69 is the sp2deg spin susceptibility ; @xcite @xmath70 and @xmath71 are the overhauser and knight shifts , respectively .
@xmath72 mev @xcite is the _ s - d _ exchange integral and @xmath73 is the thermal average spin of a single mn atom .
@xmath74 is the probability of finding an electron in the quantum well .
typical orders of magnitude in our samples of @xmath75 and @xmath76 are @xmath77 and @xmath78 ev , respectively .
interpretation of the poles is as follows . in the frame of the mn ions ,
the mn precession frequency is determined by the zeeman energy of the mn ( @xmath79 ) , the static mean - field _ s - d _ contribution from electrons ( @xmath76 ) , and the third term is the `` dynamical '' knight shift that originates from the electron precession itself responding ( through the @xmath80 response ) to the mn magnetic field ( proportional to @xmath75 ) . at @xmath40 , @xmath81 , collective modes of eq.([eqcoulomb ] ) develop two modes : @xmath82 where @xmath83 .
they open an avoided gap at a specific magnetic field @xmath84 when @xmath85 , as illustrated in fig.[figanticross ] where @xmath86 t for @xmath87 .
the upper ( resp .
lower ) energy branch is the opm ( resp . in phase ) . for @xmath88 , the opm ( resp . in phase ) has an `` electron like '' character ( resp .
`` mn - like '' ) , because it is mainly dominated by the electron ( resp .
mn ) spin precession . in resonance ( @xmath89 ) modes have a complete mixed character .
these zone center modes were observed in ref.@xcite and ref.@xcite .
the resonance field depends mainly on the mn concentration and for the experimental situation of fig.[fig : disper ] , it is about 10.7 t. figure [ fig : disper ] compares the dispersion of the opm @xmath90 , calculated using eq.([eqcoulomb ] ) and the parameters of sample a , with the experimental one . undoubtedly , the agreement with the data is excellent without any fitting procedure .
the dispersion of the sfw mode @xmath91 defined as the unique pole of @xmath92 is also plotted .
it appears that the introduction of the _ s - d _ dynamical coupling shifts @xmath90 negligibly from the pure electron mode @xmath93 for that situation .
however , the inclusion of the coulomb interaction in eq.([eqcoulomb ] ) positions the opm below the sf - spe continuum with a negative dispersion , contrary to what was found in ref.@xcite .
t and the same sample parameters as in fig.[figanticross ] . in ( a ) and ( b ) , the coulomb interaction between electrons is suppressed , only the _ s - d _ dynamical shift applies as in eq.(20 ) of ref .
the dashed line is the uncoupled mn mode energy ( @xmath94 ) . in ( c ) and
( d ) , the coulomb interaction between electrons is included as in eq.([eqcoulomb ] ) of this work .
dashed lines are the uncoupled electron mode ( @xmath93 ) and mn mode ( @xmath95 ) . at @xmath40 the sf - spe energy is shifted to @xmath96 . in ( b ) and ( d ) the _ s - d _ dynamical coupling has been artificially magnified by a factor of 100 . ]
illustration of this phenomenon is provided in fig.[figxcxsd ] . in fig .
[ figxcxsd](a ) and ( b ) the electron - mn modes dispersion were calculated using the denominator of eq.([eqcoulomb ] ) by replacing the sp2deg spin susceptibility with the non - interacting equivalent , which is single particle response [ see ref.@xcite ] .
such a calculation is equivalent to the approach of ref.@xcite .
thus , for no coulomb interaction between electrons , the pure electron modes are the sf - spe ( poles of the single - particle response ) degenerate to @xmath97 at @xmath40 .
inclusion of the _ s - d _ dynamical coupling with the mn introduces a collective opm propagating above the sf - spe continuum ( upper curve in fig . [
figxcxsd](a ) and ( b ) ) .
the relative `` blue - shift '' is labeled @xmath98 . in presence of coulomb interaction between carriers , however , a collective electron mode ( sfw ) appears and propagates below the sf - spe continuum with a negative dispersion given by @xmath91 .
the sf - spe zone center energy is `` blue shifted '' from @xmath97 to @xmath96 bye coulomb - exchange .
the relative `` red - shift '' of the sfw mode with the sf - spe is labeled @xmath99 inclusion of the _ s - d _ dynamical coupling with mn introduces a new shift which determines the final energy of the opm .
thus , we have separated two effects .
first , coulomb introduces a shift between the collective electron mode and the sf - spe energies with a relative quantity quantifiable by @xmath100 .
second , the _ s - d _ dynamical coupling with the mn shifts this mode to @xmath101 with a relative shift quantifiable as @xmath102 clearly , in the situation of fig.[fig : disper ] : @xmath103
. we may ask if conditions can be found such that @xmath104 ?
if so , the opm would propagate above the sf - spe continuum with a positive slope dispersion .
we should distinguish two situations . for @xmath105 @xmath106 in resonance , however , @xmath107 , @xmath108 calculated out or in resonance ( @xmath109 ) with the coulomb shift @xmath110 .
both are calculated as a function of the electron density for a fixed quantum well width , @xmath111 ( dotted lines ) and @xmath112 ( full lines ) .
the dynamical shift @xmath113 overcomes @xmath110 for a sufficiently low electron density ( respectively @xmath114 and @xmath115 .
the out of resonance dynamical shift @xmath98 overcomes the coulomb effect at much lower densities .
] in figure [ shiftsfig4 ] , we compare @xmath98 calculated out of resonance and @xmath113 calculated in resonance with @xmath110 for quantum well widths , @xmath111 and @xmath116 . both @xmath98 and @xmath117
are calculated as a function of the electron density , plotted in terms of @xmath118 . as seen in fig .
[ shiftsfig4 ] , the dynamical shift @xmath119 overcomes the coulomb effect at a sufficiently low electron density corresponding to @xmath114 and @xmath120 .
the out of resonance dynamical shift @xmath98 overcomes the coulomb effect at much lower densities .
( a ) for @xmath121 .
the electron modes appears above the sf - spe continuum . to get the effect ,
the mn concentration is raised to @xmath122 in order to get @xmath123 .
the external magnetic field is tuned to @xmath124 t. here @xmath112 and @xmath45 k. ] we now simulate a situation in which a prediction with @xmath125 propagating above the sf - spe is possible .
we have assumed the material parameters of fig .
[ shiftsfig4 ] , that is , when the dynamical shift dominates the coulomb shift .
we show such calculated dispersions of mixed spin modes in fig .
[ lowdensitydispersionsfig5 ] for @xmath121 . indeed ,
the electron modes appear above the sf - spe continuum .
however , to obtain such an effect , a significant spin polarization of @xmath126 must be reached by raising the mn ion concentration to @xmath122 .
this means that an external magnetic field needs to be tuned to a resonant field of @xmath124 t , an experimental challenge !
we note here that the small blue shift introduced to the sfw energy by the dynamical coupling @xmath98 , explains the near excellent agreement between our previous model and experiment . since the coulomb effect dominates in our quantum well , the mixed modes evolve essentially as an electron wave .
we have probed dispersions of spin flip waves and spin flip single particle excitations in a spin polarized electron gas in a cd@xmath0mn@xmath1te quantum well using angle - resolved raman scattering .
our key result is that the sfw dispersion lies below the sf - spe continuum , contrary to predictions made by the theory of spin waves in diluted magnetic systems .
analysis of our measurements with a model accounting for the coulomb interaction between carriers and the dynamical response of mn ions and the electron spin subsystem in the spin wave propagator of the theory agrees well with our experimental results .
we have found that the coulomb contribution dominates over the dynamical response .
we have investigated a regime in which the dynamical coupling overcomes the coulomb effect and find this to occur at an extremely large external magnetic field and for a high mn ion concentration .
the authors would like to thank jules silembo , michel menant and silbe majrab for machining parts of the magnetic cryostat and electronics used in the experiments .
special thanks to dimitri roditchev for experimental help .
this work was partially supported by the epsrc and the cnrs .
additional financial support was provided by the grant anr gospininfo .
|
we have measured dispersions of spin - flip waves and spin - flip single - particle excitations of a spin polarized two - dimensional electron gas in a cd@xmath0mn@xmath1te quantum well using resonant raman scattering .
we find the energy of the spin - flip wave to be below the spin - flip single particle excitation continuum , a contradiction to the theory of spin waves in diluted magnetic semiconductors put forth in [ phys .
rev .
b * 70 * 045205 ( 2004 ) ] .
we show that the inclusion of terms accounting for the coulomb interaction between carriers in the spin wave propagator leads to an agreement with our experimental results .
the dominant coulomb contribution leads to an overall red shift of the mixed electron - mn spin modes while the dynamical coupling between mn ions results in a small blue shift .
we provide a simulated model system which shows the reverse situation but at an extremely large magnetic field .
|
the aim of this review is to provide a snapshot of the present status of dynamical fermion simulations , both as regards their performance and the possible sources of systematic errors that are not fully under control , as well as an introduction to some of the algorithmic ideas that are currently being investigated .
as the goal is to be pedagogical rather than exhaustive we do not attempt to summarise all of the algorithmic talks that were presented at this conference . in
we consider the present state of large - scale dynamical fermion computations , comparing their costs and their control of systematic errors .
this leads us to consider the issue of the locality of using fractional powers of the fermion determinant in .
in we consider various algorithms for introducing `` fat links '' in a differentiable way . in
we review a new algorithm which allows a domain - decomposition approach to be used for lattice field theories . in
we outline the theory behind some of the algorithms that are being used to include fractional powers of the fermion determinant ; the same techniques may be applied to evade instabilities in numerical integrators ( ) , as can some other clever tricks ( ) . finally , we consider in some detail algorithms for including the dynamical effects of on - shell chiral fermions in , and in particular emphasise that all the methods used may be considered to be different approximations to the neuberger operator .
before comparing the performance and efficacy of the various dynamical fermion algorithms currently being used in large - scale numerical computations there some caveats that must be emphasised .
since such computations are extremely expensive they have only been carried out at a small set of parameter values , and thus we can not reliably compare their results or behaviour in either the continuum or thermodynamic ( infinite volume ) limit .
indeed , there are usually only results for two or fewer different lattice spacings . for the same reason it is not possible to compare the performance of the algorithms at the same physical quark masses ; for this we would need to interpolate between runs with different dynamical ( not valence ) quark masses .
some computations have been carried out with two quark flavours and others with three , and again we are forced to ignore these differences due to lack of data .
since the cost of adding in a third heavy dynamical quark is usually small compared to the cost of the two lightest quarks this probably is not too important within the large error bars .
it is very hard to make reliable estimates of the autocorrelation times involved , and thus to estimate the cost of generating statistically independent configurations from an equilibrium distribution .
not only do results depends on the operator whose integrated autocorrelations are measured , but also on the physics of the system , such as the presence of nearby phase transitions .
computations have been carried out on different lattice sizes , so we obviously need to compare the `` cost per unit volume . '' it would be wrong to assume navely that the cost of all algorithms scale linearly with the number of lattice sites @xmath0 at fixed lattice spacing @xmath1 : while the algorithm ( ) cost scales as @xmath2 for fixed ( volume - independent ) @xmath3 errors in the parameters of the effective action being simulated , the cost of the hybrid monte carlo ( hmc ) algorithm in free field hmc for a constant acceptance rate .
we shall see later ( ) that for hmc computations involving light fermions the step size @xmath4 is constrained by instability in the integrator rather than by the bulk behaviour of the hmc hamiltonian , so a @xmath2 dependence might be more appropriate for hmc too . ]
grows as @xmath5 .
fortunately the effects of such different volume extrapolations are small compared to the errors in the cost estimates anyhow .
= 0.45 -6ex shows the approximate cost of evolving a @xmath6 lattice through one unit of md time as a function of the dynamical @xmath7 mass achieved .
four types of fermion action have been used recently for large - scale computations .
_ wilson / clover _ fermions seem to increase in cost as the @xmath7 mass is reduced much more rapidly than the other formulations ; indeed , there are runs at smaller @xmath7 mass which are so expensive that they lie significantly above the graph .
it may be conjectured that this is because the dirac spectrum not bounded below , which is not the case for _ ks / staggered _ fermions ( where recent data from the milc collaboration use the improved action ) , _ twisted mass _ ( ) fermions , or _ domain wall _ ( gw / overlap ) fermions .
the question of which formulation on - shell chiral fermions is best , or indeed whether such formulations are to be preferred at all , depends on how much chiral symmetry is required . at present dynamical gw fermions
seem to be about 10100 times more expensive than the corresponding or runs at comparable masses .
the figure also indicates subjective extrapolations for the costs .
in particular it seems reasonable that the cost for dynamical domain wall fermions , or any other formulation exhibiting exact on - shell lattice chiral symmetry , should eventually become cheaper for light enough quarks because the chiral limit for them does not require taking the continuum limit .
there has been much debate recently about whether dynamical fermion formulations that weight configurations with a non - integral power of the fermion determinant correspond to local quantum field theories ( qfts ) or not , and how significant locality is .
for example , staggered quarks appear in multiples of four `` tastes ; '' when the staggered fermion fields are integrated over we are left with a fermion determinant @xmath8 in the functional integral .
this is replaced by @xmath9 to obtain a corresponding two taste theory , and gauge configurations may be generated with this weight using inexact ( ) ( ) or exact ( phmc or rhmc ) algorithms ( ) .
if a qft is local then we are guaranteed that it has the cluster decomposition property , and that within the context of renormalized perturbation theory it satisfies the familiar power counting rules , exhibits universality , and is amenable to a systematic improvement procedure . on the other hand ,
if it is not local then there is little we can say about these properties other than that we we have no _ a priori _ reason to expect them to hold . in particular , if we have a non - local lattice theory then we have no good reason to invoke power - counting arguments to justify taking the nave continuum limit , or to expect the lattice theory to be described by continuum perturbation theory however small the lattice spacing .
the fact that a formulation is not manifestly local does not logically imply that it is not local .
after all , it is easy to transform a manifestly local theory into an equivalent but non - manifestly - local form : we use this freedom to replace fermion fields with a non - manifestly - local fermion determinant , or an equally non - manifestly - local pseudofermion representation . for the case of two tastes of staggered fermions
there may be a local fermion kernel @xmath10 such that @xmath11 , however , in general a non - manifestly local theory has no reason to be equivalent to a local one .
even if there was a local action corresponding to taking fractional powers of the fermion determinant in the functional integral , we are still required to use this local action to measure fermionic quantities . for the case of staggered quarks
this means one must use the hypothetical @xmath10 rather than @xmath12 for valence measurements .
we should not expect that measuring operators corresponding to a local four taste valence action @xmath12 on configurations generated with @xmath13 to lead to consistent results . not only might there be unknown renormalisations of the parameters between the sea and valence actions ( e.g. , what is the justification for using the same numerical value for the quark masses ? ) but the degrees of freedom are not even the same . unless we know @xmath10 explicitly or are only interested in measuring purely gluonic observables we are forced into having a `` mixed action , '' that is different sea and valence actions , and such a situation suffers from the same problems , such as violation of unitarity , as quenched qcd .
of course , one might hope that these problems do not occur , or at least are smaller than for an arbitrary mixed action situation , but these systematic errors are not quantitatively under control .
some recent publications @xcite have studied the question as to whether the operator @xmath14 is in fact local . to make this question meaningful we need to specify which square root is under consideration .
if the square root of the lattice operator is defined by taking the square roots of its eigenvalues in a basis in which it is diagonal then the sign of these roots is arbitrary . in the tests that have been carried out
the square root has been defined by a polynomial approximation , which corresponds to choosing the positive square roots .
suppose we applied this procedure to the wilson dirac operator @xmath15 ; we would find that the square root was non - local , because the local square root ( @xmath16 ) is not a positive operator .
it is important to stress that the fact that @xmath14 appears to be non - local does not prove that there is no corresponding local operator @xmath10 with the same determinant ; on the other hand the only operator we know with this determinant ( and with staggered lattice symmetry ) is @xmath14 itself , so it is worthwhile verifying that it is not ( miraculously ) local as has sometimes been suggested as a possibility .
after all , the neuberger operator appears to be local on sufficiently smooth gauge configurations @xcite , despite the fact that it is not manifestly local .
hoping for the existence of @xmath10 , or even proving it , is insufficient : we would need to use this operator for valence measurements in order to have a manifestly consistent unitary theory .
let us now briefly review how the locality of a lattice dirac operator may be measured numerically .
we define a `` wavefunction '' by applying a lattice dirac operator to a point source , @xmath17 .
we then say that the corresponding qft is ultralocal if the wavefunction has `` compact support , '' @xmath18 , and is local if the wavefunction has `` fast decrease , '' @xmath19 , with @xmath20 being the localisation length .
note that in both cases the wavefunction falls to zero at any non - zero physical distance in the continuum limit , thus corresponding to a local continuum theory .
= 0.49 -6ex typical results for the effective localisation length @xmath21\right\}^{-1}$ ] are shown in as a function of the physical distance ( @xmath22 is the sommer scale ) .
the effective localisation length does not reach a plateau in the region where we may hope that finite - size effects are under control , so we can not extract a definite estimate of @xmath20 , on the other hand the evidence is that it must be non - zero at a non - zero physical distance , independent of @xmath1 , so the operator can not be local in the continuum limit .
furthermore , the magnitude of the localisation length is comparable with the inverse @xmath7 mass , and thus use of @xmath14 would not lead to meaningful ( valence ) spectrum measurements .
the reader is also referred to interesting related work in the schwinger model @xcite .
we shall now survey the topic of `` fat links . ''
the idea of replacing gauge link variables by some average of their neighbours is not new , and for a long time has been used to construct operators that are good sources and sinks , in the sense that they have a better overlap with ground state .
the basic idea is that instead of explicitly constructing a more complicated less local improved operator we use the simplest operator with the correct symmetries ( quantum numbers ) but taken as a function of the smeared or fattened link fields .
the fattening procedure is thus equivalent to building a physical size smeared source . of course , to calculate masses and so forth we need to measure temporal correlation functions , and the sources and sinks for these should be localised in time , so we typically want to use spatial link fattening only in this case .
another use of fat links which is becoming important as dynamical fermion computations become commonplace is that of smearing the action itself , in this case by computing the original action as a function of the fattened link fields . here
the goal is to suppress uv fluctuations in the gauge fields .
this may be considered as a means of constructing improved actions in the spirit of renormalisation - group improved ( or `` perfect '' ) actions , or systematic improvement in powers of @xmath1 ( either perturbative or not ) .
there have recently been two main applications of fat links : improving gauge actions so as to produce smoother gauge fields upon which quenched gw dirac operators are localised , and improving fermionic actions themselves .
we shall consider the latter here , as we are concerned with dynamical fermion algorithms .
the problem with using a highly improved action such as hyp @xcite or perfect actions @xcite in dynamical computations is that we either need to find non - md - based algorithms for generating the gauge configurations including the effect of the fermion determinant , or we need to compute the md force corresponding to the improved actions .
the most successful fat actions , such as hyp , used in quenched computations have been those where the fat links are somehow projected onto the group manifold . indeed
, this seems to be a crucial ingredient for their success .
unfortunately this projection would seem to be a non - differentiable operation ( i.e. , the fat links are not a smooth function of the original thin links ) . the first attempt to show that this was not necessarily so , and indeed that the md equations of motion for fat actions could be practicable , was using the flic ( fat link improved clover ) action @xcite .
let @xmath2 be the usual unprojected ape - smeared link , then the usual iterative projection @xmath23 where @xmath24 maximizes @xmath25 is not differentiable ; however by defining successively @xmath26 and @xmath27 instead we obtain a mapping that is almost differentiable . the `` almost ''
is that differentiability fails along the branch cut of the cube root , but this does not seem too important in practice .
= 0.4 another `` projection '' procedure which is truly differentiable is that of `` stout links '' @xcite .
let @xmath2 be a suitably smeared link and @xmath28 the corresponding thin link , so that @xmath29 is an untraced sum of plaquettes .
this may be `` projected '' onto @xmath30 by first projecting it onto the @xmath31 algebra and then exponentiating it into the group , @xmath32 , where @xmath33 means the traceless antihermitian part .
this is now differentiable , but it does not look too much like projection ( except when @xmath34 ) . in quenched tests
stout links seem to be about as good as ordinary projected fat links , but require a little more tuning .
these methods can be applied iteratively to produce differentiable links of arbitrary obesity .
an interesting new algorithm based on the schwarz alternating procedure ( sap ) has been introduced by both as a preconditioner for solving the lattice dirac equation @xcite and as dynamical fermion monte carlo algorithm @xcite .
the sap , introduced in 1870 , was probably the first domain decomposition method for solving the dirichlet problem for elliptic partial differential equations in complicated domains .
the inequalities used in the convergence proof of the sap for pdes rely on the elliptic nature of the systems .
= 0.5 -6ex the basic idea is to decompose the lattice into blocks , and then update each block with boundary conditions specified by the most recent fields on the neighbouring blocks , as illustrated in . in the figure @xmath35 blocks
are shown , but when considering the scaling behaviour of the algorithm the blocks should be taken to be of fixed physical size .
when we invert the dirac operator on the lattice we modify the ( pseudo)fermion fields in a fixed gauge background . for a nearest - neighbour fermion
kernel , such as the wilson dirac operator , the fermion fields in the even blocks are coupled only to those in other odd blocks and vice versa .
we can thus alternate updates of all the even and all the odd block fields , giving a procedure that has a large degree of parallelism .
notice that we are not just alternating single updates but we are alternating complete solutions of the dirichlet problem on each set of blocks .
this procedure converges to the correct solution , but the rate of convergence is slow ; so instead of using sap as a solver we may use it as a preconditioner for a space solver ( such as gcr or fgmres ) . for this purpose
accurate block solves are not required , as the preconditioner need only be an approximation to the inverse , and only a few schwarz cycles are required .
this method seems to lead to a significant speed - up by a factor of 23 over a range of quark masses , as illustrated in .
the algorithm parallelises easily , especially on coarse - grained architectures such as pc clusters .
= 0.5 -6ex a more radical use of the sap is as a dynamical fermion algorithm itself , in which we update the gauge fields in the blocks .
the pseudofermion term in the action involves the inverse of the lattice dirac operator , and is thus manifestly non - local . in order to allow the update of a subset of the gauge links we make use of the schur complement factorisation of the quark determinant : @xmath36 .
we indicate the grey and white block of by @xmath37 and @xmath38 respectively , and exterior boundaries ( indicated by white dots in the figure ) by @xmath39 and @xmath40 . this decomposition is easily derived from the ul factorisation @xmath41 we may now introduce two pseudofermion fields , @xmath42 with support on the interior of the blocks and @xmath43 with support on the boundary ; the associated pseudofermion part of the action being @xmath44 , with @xmath45 being the schur complement operator introduced above .
the gauge fields on the `` active '' links within the blocks together with the @xmath42 pseudofermions can be equilibrated with @xmath43 held fixed , and then the effect of the `` schur complement '' interaction @xmath45 can be incorporated into a global hmc process .
we occasionally shift the blocks to ensure all links get their turn to be active and thus updated . according to the preliminary results that have been presented
, this scheme successfully separates the short- and long - distance effects . within the blocks
the smallest eigenvalues of the dirac operator are regulated by the inverse block size , whereas for the force due to the @xmath43 pseudofermions is small and only weakly mass dependent in the interior of the blocks ; this allows larger step sizes for the global hmc process .
this looks very promising , and it will be interesting to see how the costs scale for very small quark masses . so far
the procedure has been applied to wilson fermions , and it might be painful to generalise it for those fatter fermion actions for which the fraction of `` active '' links becomes smaller .
previously in we discussed the validity of using a fractional power of the fermion determinant in order to circumvent the doubling problem for staggered fermions . in this section
we shall ignore such matters of principal and consider the available repertoire of practical dynamical fermion algorithms available for this case .
to date , all large - scale computations with two or three tastes of staggered quarks have used the algorithm @xcite .
this is an inexact algorithm , in the sense that it is a markov process that converges to an equilibrium distribution that only approximates the desired one .
it is based on the alternation of md trajectories and gaussian momentum refreshment , but with no accept / reject step to correct for non - zero step size errors . by using a clever combination of non - reversibility and area non - preservation in the md integrator , the algorithm converges to a distribution with errors of @xmath3 in the parameters of the action . as already noted in these @xmath3 errors
are volume independent .
it is important to note that this result is an asymptotic expansion in @xmath4 , so there are also errors of the form @xmath46 which are only small if the quantity @xmath47 where @xmath48 is some suitable scale parameter .
indeed , we may expect these non - analytic errors to become dominant at the same @xmath4 for which the md becomes unstable ( ) ; we should not expect to be able to use a larger step size than for the corresponding exact hmc algorithms , except that instead of getting a tiny acceptance rate we would just quietly get an incorrect answer .
the observation that the estimate of the errors in the equilibrium distribution is an asymptotic expansion .
indeed , we may reasonably expect scaling for more highly - improved theories to break down at smaller values of @xmath1 .
] tells us that we have no reason to expect the results for large @xmath4 to just correspond to a renormalisation of the parameters , as was originally suggested for the related langevin algorithm .
relatively recently the polynomial @xcite and rational @xcite hybrid monte carlo algorithms have been developed .
these permit the simulation of fractional powers of the fermion determinant without any concomitant systematic errors in the equilibrium distribution .
the basic idea is to use a polynomial or rational function approximation to the power occurring in the action . recall that in the context of the hmc algorithm an approximate action suffices for md evolution , and an accurate action is only needed for the acceptance test .
this permits us to use a relatively cheap approximation for each md step , and a more expensive approximation for the infrequent acceptance tests .
the fermion kernel @xmath12 on the lattice is a large sparse matrix , and we wish to use its inverse square root , or more generally some continuous function @xmath49 of it .
we define such functions on hermitian matrices by diagonalisation , if @xmath50 where @xmath51 is diagonal then @xmath52 .
the key property of polynomial and rational functions is that they do not require the diagonalisation to be carried out explicitly , since @xmath53 and @xmath54 .
the theory of optimal @xmath55 ( ) approximation over a compact interval is well understood : s theorem tells us that there is a unique optimal approximation characterised by having alternating error maxima of equal magnitude ( c.f . , ) . in general
we can determine the coefficients of the optimal approximation numerically using algorithm , but for the interesting cases of @xmath56 and @xmath57 they are have been calculated in closed form in terms of elliptic functions by .
= 0.45 -6ex it is not necessary to use the optimal approximation , but doing so leads to surprisingly small errors ; for example , if we consider rational approximations to @xmath56 of degree @xmath58 then the optimal / approximation has a maximum error of @xmath59 over @xmath60 , clearly has a maximum error of about 1 .
it is worth emphasising the rle played by polynomials in the theory of optimal approximation .
these polynomials are defined by @xmath61 , and have exactly @xmath62 alternating extrema of unit magnitude over the interval @xmath63 .
this means that @xmath64 is the best approximation to @xmath65 of degree @xmath66.$ ] with respect to the weight @xmath67 , up to a trivial factor in the normalisation of @xmath68 . nevertheless , truncated expansions are not optimal in general , even for polynomials . ]
there are some advantages of rational over polynomial approximations . in most cases
the error for approximations of a given degree are much smaller for rational approximations ; for example for the @xmath69 $ ] error for @xmath70 is proportional to @xmath71 for the best rational approximation of degree @xmath72 whereas the best polynomial approximation has an error that only falls as @xmath73 .
the rational approximation s exponential convergence means that it can be made exact to machine precision in a numerically stable and cheap manner .
the obvious advantage of a polynomial approximation is that it does not require the numerical solution of systems of linear equations , but bear in mind that the polynomial approximation to @xmath74 corresponds to computing the matrix inverse using the method , as compared with the more efficient solvers that are used in the computation of rational approximations .
if the rational functions are expressed as a partial fraction expansion , and a multi - shift solver @xcite is used to compute all the necessary inverses simultaneously in the same space , then rational approximations are very competitive .
in order to understand some of recent techniques for decreasing the cost of dynamical fermion hmc computations by increasing the integration step size we need to briefly review some earlier results on the mechanisms which limit these step sizes @xcite .
= 0.46 -6ex in order to satisfy detailed balance and thus be an exact algorithm we need to use an area - preserving reversible md integration scheme within the hmc algorithm .
fortunately symmetric symplectic integrators , of which the familiar leapfrog scheme is the lowest - order example , have these desirable properties .
they are only exactly reversible up to floating - point rounding errors , and these errors become important when they are exponentially amplified by the chaotic nature of the underlying equations of motion @xcite .
this is illustrated in , where the exponent @xmath75 that characterizes this chaotic amplification is plotted as a function of the integration step size @xmath4 .
notice that @xmath76 in all three graphs ; this corresponds to the fact the underlying continuous time equations of motion are indeed chaotic .
this interpretation is supported by the fact that the @xmath75 seems to have a constant value independent of @xmath4 .
when @xmath4 exceeds some critical value @xmath77 then @xmath78 ; this corresponds to an instability of the leapfrog integrator , a phenomenon which occurs for the leapfrog scheme even for a single harmonic oscillator .
@xmath77 depends on quark mass : for the quenched theory or the case of heavy fermions the instability occurs at values of @xmath4 so large that the acceptance rate is already essentially zero , since @xmath79 ; with light fermions the instability sets in for much smaller values of @xmath4 , and is what prevents larger step sizes from being used . indeed ,
this is the reason why the use of higher - order symmetric symplectic integrators is not useful ; they increase the exponent @xmath72 characterising the growth of the bulk contribution , @xmath80 , but yet again this is only an asymptotic expansion and the contributions of the form @xmath81 dominate when @xmath82 . in order to include the dynamical effects of fermions we need to evaluate a functional integral including the fermionic determinant @xmath8 .
the usual method of doing so is to write this as a bosonic functional integral over a pseudofermion field with kernel @xmath83 , using the identity @xmath84 .
we then alternate markov steps of ( i ) selecting a pseudofermion field from a gaussian heatbath with the gauge fields @xmath85 held fixed , and of ( ii ) updating the gauge fields , e.g. , by using an hmc algorithm , while keeping the pseudofermion fields fixed .
this is a perfectly valid procedure , but we are introducing extra noise into the system by only using a single pseudofermion field to sample the functional integral .
this noise manifests itself as fluctuations in the force exerted by the pseudofermions on the gauge fields ; this increases the maximum fermion force , which may trigger the integrator instability ; and this in turn requires us to decrease the integration step size .
a less noisy estimate is to use the identity @xmath86^n $ ] to introduce @xmath72 _ multipseudofermion _ fields , @xmath87 this should increase the value of @xmath77 . in the limit of an infinite number of multipseudofermions the force on the gauge fields
is computed `` exactly , '' but this is not the same as computing the force due to @xmath8 , because the pseudofermions are updated at beginning of a gauge field trajectory , and do not represent @xmath8 except at the beginning of the trajectory .
it is not at all clear that this is in any way a problem , however .
alternating gauge field trajectories with pseudofermion heatbath updates is a valid markov chain with the correct equilibrium distribution .
furthermore , the md ensures that the gauge fields evolve along a trajectory in the fixed pseudofermion background , and this is not the same trajectory as they would take under the `` instantaneous '' force due to @xmath8 .
there appears to be no reason for the pseudofermion force along a trajectory to be larger for one sort of trajectory that for the other : they are just different markov chains with the same fixed point distribution .
the first method of implementing multipseudofermions was introduced by hasenbusch @xcite . for the case of the wilson fermion action
@xmath88 he introduced the quantity @xmath89 , and used the identity @xmath90 to write the fermion determinant as @xmath91 .
he then used a separate pseudofermion for each determinant , and tuned @xmath92 to minimise the cost .
this method easily generalises to more than two pseudofermions , and to more complicated actions such as the wilson - clover action .
another way of implementing multipseudofermions is to use the rhmc technique ( ) to implement @xmath72th root of the fermion kernel @xcite .
following the usual procedure for rhmc we use a partial fraction expansion of the optimal rational approximation for @xmath72th root , and use a multishift solver to compute all the terms in the same space .
the advantage of this approach is the no tuning is required , as the condition number @xmath93 is divided equally among all the multipseudofermions . if we make the nave assumption that the cost is proportional to condition number of the matrices we need to invert , then the condition number for each multipseudofermion kernel is @xmath94 , so the total fermionic force is reduced by factor @xmath95 . if this permits us to increase the step size by the same factor then the optimal value @xmath96 minimises the cost , with an optimal cost reduction of @xmath97 .
of course , this argument is too nave , we can not reduce force by an arbitrarily large factor just by increasing the number of multipseudofermion fields @xmath72 . at some point the fluctuations in the force coming from the stochastic estimate of the fermion determinant are no longer larger than the force itself , at which point increasing @xmath72 much further will not help .
= 0.25 -7ex an interesting `` replay '' trick for reducing fluctuations in @xmath98 along an hmc md trajectory has been suggested by and sommer @xcite .
consider the situation shown in .
the system starts at point @xmath99 in phase space , and the energy change along the trajectory to @xmath100 with step size @xmath4 is @xmath101 .
if @xmath101 is small , say @xmath102 , then we accept the new configuration with probability @xmath103 or else keep the old configuration @xmath28 ; but if @xmath104 we construct a more accurate trajectory from @xmath105 with step size @xmath106 . in order to ensure reversibility we need to make sure that this same reduced step size would be used if we started at @xmath107 , so we must verify that the fictitious energy change along the trajectory from @xmath108 with step size @xmath4 has energy change @xmath109 .
if this is the case we accept @xmath110 with probability @xmath111 or else keep @xmath28 , and everything is area - preserving and reversible .
if @xmath112 then we would not try reducing the step size in the reverse direction , so we fall back to accepting @xmath28 . by considering the cases where the energy changes @xmath101 , @xmath113 , and @xmath114 are negative , between 0 and 1 , and @xmath115 separately detailed balance is easily verified .
this procedure is very useful if the system occasionally hits a large energy fluctuation , for example if it comes across an integrator instability .
if such instabilities occur frequently then we must ask whether it would not be preferable just to reduce the step size ( say to @xmath106 ) for all trajectories . another issue to consider
is that often these large @xmath98 changes occur as the system is approaching equilibrium , and we will be forced to reduce @xmath4 anyhow as the system becomes more equilibrated and `` sees '' the large force due to the light fermions .
another interesting trick that is sometimes useful was introduced by sexton and weingarten @xcite over a decade ago .
their idea is to split md hamiltonian into two parts , usually corresponding to the boson and fermion parts of the action , and to construct a symmetric symplectic integrator with larger steps for more expensive ( fermion ) part by a clever application of the baker
campbell hausdorff formula .
this helps if the step size is limited by cheaper ( boson ) part , but unfortunately becomes less useful as @xmath116 when the step size is limited by the integrator instabilities induced by the fermion force .
nevertheless , it is a useful technique when the large contributions to the force are not also the most expensive .
we now turn to one of the most important developments of recent years : the formulation of on - shell chirally symmetric fermions on the lattice .
our goal is to show that the various techniques developed for this purpose , such as the domain wall or overlap formulations , are equivalent , in the sense that they are just different approximations to the same on - shell chiral lattice theory .
there are many variants of the methods , but they all seem to correspond to choosing ( explicitly or implicitly ) a rational approximation to the @xmath117 function , and of computing the inverse of the corresponding approximate neuberger operator .
let us fix our notation and conventions : we shall work in euclidean space with hermitian @xmath118 matrices ; and we shall take all dirac operators to be @xmath119 hermitian , @xmath120 .
although it was not the way the subject developed historically ( a matter that we will not even attempt to review here ) , an elegant logical approach @xcite is to ask if it is possible to have chiral symmetry on the lattice without doublers if we only insist that the symmetry holds on shell .
such a transformation would have to be of the form @xmath121 which becomes a chiral transformation on - shell where @xmath122 .
note that we are free to specify the chiral transformation properties of @xmath123 and @xmath124 independently .
for this transformation to be a symmetry the dirac operator must be invariant @xmath125 for a small transformation , @xmath126 , this implies that @xmath127 , which is just the ginsparg wilson ( gw ) relation @xmath128 . we can find a solution of the ginsparg
wilson relation as follows : let the lattice dirac operator be of the form @xmath129 , where @xmath130 , thus @xmath131 .
this satisfies the ginsparg
wilson relation if @xmath132 , and it has the correct continuum limit @xmath133 if @xmath134 .
both of these conditions are satisfied by the choice @xmath135.\ ] ] there are many other possible solutions , but these essentially all correspond to just using a different dirac operator within neuberger s operator
. we should emphasise that we are only considering vector - like theories with a chiral symmetry here , chiral theories with unpaired weyl fermions can be discretised on the lattice , but getting the phase of the fermion measure correct is critical . simulating such theories is a much harder problem than the already challenging computations required for the case considered here .
all numerical techniques for calculating with on - shell chiral fermions may be viewed as different ways of approximating neuberger s operator .
it is relatively straightforward if somewhat expensive to apply the operator , for instance by using a rational or polynomial approximation to the @xmath117 function , but it is more challenging to apply its inverse .
the difficulty is that neuberger s operator is constructed out of two non - commuting operators , @xmath119 and @xmath136 , so the inverse of a rational approximation is not just the reciprocal rational function .
there are various ways the inverse may be computed : the most direct approach is by a nested space approach , where for every outer iteration a full inner inversion is required to apply the neuberger operator .
the obvious problems with this approach are ( i ) the information built up in the inner space is not made use of in the outer space ; ( ii ) the accuracy ( residual ) required for the inner solves in order to reach a specified accuracy for the outer solve is not immediately obvious .
quite a lot of work has gone into developing these techniques , including recent studies of various preconditioners and the use of `` flexible inverters '' for which the approximation used for the preconditioner may vary during the outer inversion @xcite .
an alternative approach to computing the inverse of the neuberger operator is to express it as a single inverse of a five dimensional operator .
this may done is by linearising a continued fraction representation of a rational approximation to the neuberger operator @xcite by the introduction of a set of @xmath137 auxiliary spinor fields ; these can be introduced either as constraints or as new dynamical variables . from this viewpoint
the new `` fifth '' dimension is just an means of approximating the @xmath117 function , and the gauge fields are still only four dimensional .
this is similar to the domain wall formulation @xcite , and it is possible to translate domain wall fermions into an equivalent four - dimensional approximation to the neuberger operator @xcite . to do this
we first separate the left- and right - handed parts of the domain wall kernel and cyclically shift the latter by one site in the fifth dimension . in the domain wall formalism
the four - dimensional fermion fields have their left - handed components on one wall and their right - handed components on the other wall , and the shift moves them both to the same wall , as well as reducing the kernel to a form in which we can identify an explicit fifth - dimensional transfer matrix @xmath33 . in this form
we can integrate out the bulk fermion fields , cancelling the jacobian with the contribution from the five - dimensional pseudofermion fields , and construct an effective four dimensional hamiltonian @xmath138 , which is the euclidean form of a cayley transform of the transfer matrix .
this gives a higham approximation @xmath139 to the neuberger operator @xmath140 .
borii also introduced the `` truncated overlap '' for which @xmath141 , and chiu has shown how to modify the domain wall formalism to give a approximation in the equivalent effective four - dimensional theory @xcite . at this meeting brower ,
originos , & neff @xcite introduced `` mbius fermions , '' which interpolate between domain wall and truncated overlap fermions .
we see that there are many formulations of overlap fermions , all of which satisfy the gw relation , and which are equivalent in continuum limit .
the principal differences between them are that they use different lattice dirac operators within the @xmath117 function , and different approximations to the @xmath117 function .
this is presumably also true for perfect actions .
the choice between the formulations is essentially just a trade - off between computational speed and the amount of chiral symmetry breaking . to date
, most applications of on - shell chiral fermions have been for valence computations , but of course we need to introduce dynamical ( sea ) overlap fermions too in order to be able to obtain reliable chiral results from full qcd .
the first studies of dynamical overlap were carried out in the schwinger model @xcite several years ago .
large - scale dynamical domain - wall qcd computations have been carried out by the rbc collaboration @xcite .
recently fodor , katz , & szab @xcite and cundy , krieg , frommer , lippert , & schilling @xcite have studied dynamical overlap fermions in qcd on `` ridiculously small lattices . '' apart from the issues in common with valence overlap computations that we have discussed previously , hmc computations involving overlap fermions require the computation of the md force . in particular , there is an issue concerning what to do when one of the eigenvalues of @xmath16 passes through zero , which for continuous time md evolution with an exact neuberger operator would lead to a singular force .
if a polynomial or rational approximation to the @xmath117 function is used then this is readily differentiable and leads to a computable force term , so one method of handling the singular force is just to observe that there is no discontinuity in the approximations to the @xmath117 function .
this is effectively the method used in the dynamical domain wall computations , where the higham approximation is implicitly used and which , for practicable values of @xmath137 , smoothes the discontinuity enough to keep the force under control .
of course , this is just a reflection of the fact that chiral symmetry is not very well approximated .
another approach @xcite is to modify the leapfrog integrator to reflect or refract at a zero crossing in such a way that the md process is still exactly reversible and area preserving
. it will be interesting which , if any , of these approaches avoids the hmc acceptance rate becoming small due to zero crossings .
we also note the suggestion @xcite of working in the chiral sector with no zero modes , and explicitly reweighting the configurations with @xmath142 where @xmath143 is the topological charge .
we are at an interesting point in the development of algorithms for dynamical fermion computations in qcd . on the one hand
it is now generally accepted that quenching leads to significant systematic errors , and that to proceed towards reliable higher accuracy calculations we need to include dynamical fermions one way or another . on the other hand
, there is much debate about whether the systematic errors of two or three flavour staggered quark simulations are under control .
recent progress with dynamical and gw fermions also show that the prospects for carrying out large - scale dynamical computations with light quarks , with all sources of systematic errors under control , and even with good chiral symmetry properties should be expected with the current or next generation of computers .
i would like to thank martin and urs wenger for making helpful comments on draft versions of this article .
|
we consider recent progress in algorithms for generating gauge field configurations that include the dynamical effects of light fermions .
we survey what has been achieved in recent state - of - the - art computations , and examine the trade - offs between performance and control of systematic errors .
we briefly review the use of polynomial and rational approximations in hybrid monte carlo algorithms , and some of the theory of on - shell chiral fermions on the lattice .
this provides a theoretical framework within which we compare algorithmic alternatives for their implementation ; and again we examine the trade - offs between speed and error control .
|
in a recent paper @xcite , smirnov & smirnov have studied the scattering of two - dimensional ( 2d ) vortex - antivortex pairs and solitons by a single quantum vortex in a homogeneous atomic bose - einstein condensate .
they found that the pair is scattered over large angles radiating sound waves , in agreement with earlier calculations @xcite .
this scattering process is important because it lies at the heart of the dynamics of 2d quantum turbulence , a problem which is currently attracting experimental and theoretical attention @xcite .
our understanding of the turbulent motion of many interacting vortices is based on recognizing the most elementary interactions , such as the interaction of a vortex with another vortex of the same or opposite sign ( resulting respectively in rotational or translation motion of the pair ) .
similary , we would like to recognize the possible elementary interactions between a vortex and a large density perturbation induced by the dynamics of vortices by external means .
it is well - known that a quantum vortex in a bose - einstein condensate is a hole of zero density around which the phase changes by @xmath0 .
the natural question is whether the incoming vortex - antivortex pair would be scattered ( and if so , by which amount ) by a density pertubation alone ( without the circulation around it ) , as density gradients induce a magnus force @xcite which deflects the pair . to answer this question ,
we have performed numerical simulations of vortex - antivortex pairs travelling towards a fixed target in the form of a density perturbation ( hereafter referred to as an ` impurity ' ) and whose depth and size is similar to the depth and size of a quantum vortex ( but without the circulation ) . here
we report about the significant scattering induced by the impurity , and compare it with the scattering induced by a target in the form of a vortex . for simplicity
we consider a homogeneous condensate at zero temperature , and aim at identifying the various qualitative scenarios which are possible ( quantitative predictions of vortex trajectories in a harmonically trapped condensate require more specific calculations which depend on the actual physical parameters and geometry , and are outside the scope of this work )
. a better physical understanding of the scattering which imperfections induce on vortices is generally useful ( although in general imperfections may not be as symmetric as we describe them here ) .
our results are also relevant to the manipulation of vortices using optical potentials generated by laser beams @xcite .
our model is the 2d gross - pitaevskii equation ( gpe ) for a homogeneous condensate at zero temperature .
we use dimensionless variables based on the healing length @xmath1 , the time scale @xmath2 and the number density @xmath3 , where @xmath4 is the chemical potential , @xmath5 is the ( 2d ) interaction parameter , @xmath6 and @xmath7 is planck s constant ; the external potential @xmath8 is scaled by @xmath4 .
the resulting dimensionless gpe @xmath9 is solved in the ( dimensionless ) 2d domain @xmath10 @xcite .
the initial condition at @xmath11 , schematically described in fig .
[ fig1 ] , is a vortex - antivortex pair , consisting of a left ( clockwise ) vortex and a right ( anticlockwise ) vortex initially placed respectively at positions @xmath12 , @xmath13 and @xmath14 , @xmath15 .
we call @xmath16 . the initial distance between the vortices of the pair .
the vortex - antivortex pair travels along the negative @xmath17 direction with impact parameter @xmath18 towards a fixed density perturbation ( or impurity ) held at @xmath19 .
the impurity is represented by the ( dimensionless ) external potential @xmath20 where @xmath21 . with a suitable choice of parameters @xmath22 @xcite , solving the time - independent gpe without the vortex - antivortex pair ,
the density profile of the impurity approximately matches the profile of a singly - charged vortex at @xmath23 , @xmath24 in the homogeneous condensate . to quantify the scattering
, we measure the deflection angle @xmath25 of the antivortex away from its initial trajectory , but in some cases ( for example if the vortex - antivortex pair breaks up ) a different description of the interaction is necessary .
in all calculations we choose initial y - coordinates @xmath26 , sufficiently away from the impurity ; the initial x - coordinates , @xmath12 and @xmath14 , vary from case to case , as we change the impact parameter @xmath7 and the vortex separation @xmath27 .
we have identified three typical scenarios : 1 . * fly - by scenario * + if the impact parameter @xmath7 is large and negative , the vortex - antivortex pair is too far at the left of the impurity , see fig .
[ fig2](a ) , to be affected , and the deflection angle is @xmath28 . if @xmath7 increases ( still keeping @xmath29 ) , the vortex - antivortex pair is scattered to the left with increasing positive deflection angle @xmath25 , as shown in fig .
[ fig2](a , b ) .
2 . * trapping scenario * + if @xmath7 is further increased ( still keeping @xmath29 ) , the vortex falls into the region of low density of the impurity , see the red trajectory of fig .
[ fig2](c ) , becomes trapped and stops , with a strong emission of sound waves , see fig .
[ fig3 ] ; at this point , the isolated antivortex processes around the impurity , see the blue trajectory of fig .
[ fig2](c ) .
clearly in this scenario the deflection angle can not be defined .
* go - around scenario * + a further increase of @xmath7 means that the vortex - antivortex pair is almost aimed at the impurity ; the left ( anticlockwise ) vortex and the right ( clockwise ) vortex overtake the impurity on opposite sides , going around it along opposite directions , before joining again , re - forming the pair , and moving on to infinity .
[ fig2](d ) shows that for slightly negative values of @xmath7 the vortex pair is scattered to the right ( @xmath30 ) ; for @xmath31 , see fig .
[ fig2](e ) , the vortex - antivortex pair proceeds almost straight ( @xmath28 ) , vortex and antivortex going around the impurity in opposite directions .
finally , for larger , positive values of @xmath7 , the trajectories of the vortex and the antivortex are the same ( as the impurity does not introduce any preferred orientation ) , @xmath25 being replaced by @xmath32 ( in other words the function @xmath33 is antisymmetric in @xmath7 ) .
we summarize the scenarios which we have revealed by plotting the deflection angle @xmath25 as a function of the impact parameter @xmath7 , see the red line and dots in fig .
[ fig4](top ) .
the shaded areas represent the regions where the deflection angle @xmath25 can not be defined because one vortex becomes trapped .
it is instructive to replace the impurity with a third vortex , choosing positive anticlockwise circulation , initially placed at @xmath19 . in this way we can directly compare the deflections of the vortex pair s trajectory caused by a third vortex to the deflection caused by an impurity with the same density perturbation , isolating the effect of the vortex circulation . unlike the impurity , which is fixed ,
the third vortex is free to move under the velocity field of the vortex - antivortex pair .
the deflection angle @xmath25 caused by the third vortex is shown by the blue line and dots of fig .
[ fig4](top ) .
it is apparent that , for large negative impact parameters , the deflection angle @xmath25 is approximately the same for vortex and impurity , but becomes significantly larger for the vortex at small @xmath7 ; moreover , there is no trapping regime for the vortex .
note also that , for the vortex , the curve @xmath33 is not antisymmetric , as in the case of the impurity : for @xmath29 , the closest interaction is between vortices of the same sign , which makes the vortices to rotate around each other causing a deflection to the left ( @xmath34 ) ; for @xmath35 , the closest interaction is between vortices of the opposite sign , which makes the vortices to travel away together , causing a deflection to the right ( @xmath36 ) .
finally , notice that the effect of the third vortex extends to large positive values of @xmath7 , unlike the impurity .
[ fig4](bottom ) shows what happens if we halve the initial separation of the vortex from the antivortex to @xmath37 . at this short separation ,
the trapping regimes disappears ( the vortex , now rather close to the antivortex , moves at large speed , and the impurity is not strong enough to stop it ) .
the other features of the interaction remain qualitatively the same as for @xmath38 . increasing the size of the impurity or making it shallower - as for the density profiles shown by the dashed red and blue lines of fig .
[ fig5 ] - does not change the scenarios of interaction with the vortex - antivortex pair in a qualitative way . fig . [
fig6](top left ) and fig .
[ fig6](bottom left ) show the go - around scenario respectively for a larger impurity and for a shallower impurity ( the density at the centre is only @xmath39 ) .
it is interesting to notice that phase defects ( ghost vortices ) appear inside the large impurity - compare the top right and bottom right panels of figs .
we have compared trajectories of vortex - antivortex pairs launched either towards a third vortex or toward an impurity in the form of a similar density hole but without the circulation . by varying the impact parameter ,
we have identified three general scenarios ( fly - by , trapping , go - around ) which can occur . in the first scenario , the effect of the impurity if qualitative similar to that of the vortex , in the second and third scenarios it is significantly different .
these scenarios represent the elementary processes which can be recognized within a turbulent system .
they are therefore relevant to experiments in which vortices are manipulated by laser beams and to studies of 2d quantum turbulence , as large density perturbations are often generated by vortex annihilations or by the moving laser beam used to nucleate vortices in the first place .
future work should look at the effects induced by an inhomogeneous density background in harmonically trapped condensates .
other aspects which are worth investigating are thermal and quantum fluctuations , which are not included in our mean field gpe model .
qualitatively , one would expect thermal fluctuations to move the vortex and the antivortex of a pair closer to each other , eventually leading to their annihilation .
this effect would lead to the introduction of a new length / time scale associated with the vortex - antivortex pair s intrinsic decaying dynamics .
qualitatively , however , we would still expect the same regimes to emerge .
quantum fluctuations would lead to an intrinsic jitter motion of each vortex about its mean position , with the target impurity also suffering some fluctuations from the fluctuating density in that region . on the average we would still expect the fly - be and go - around scenarios to persist , but our discussion here may represent a rather idealised case .
in fact it would be interesting to investigate this scenario experimentally .
we choose @xmath40 and impose @xmath41 on the boundaries .
typically we use a @xmath42 grid , corresponding to the ( dimensionless ) spatial discretization @xmath43 .
time - stepping is performed using the 4th - order runge - kutta scheme ; the typical ( dimensionelss ) time step is @xmath44 .
we have checked that during typical evolutions the total energy is conserved within @xmath45 .
the initial vortex configuration is integrated in imaginary time for ( typically ) 3500 time steps ( enforcing the normalization @xmath46 ) in order to achieve the lowest energy state , and then integrated in real time to determine the trajectory of the vortex - antivortex pair . ; the impact parameter is @xmath7 ( here @xmath29 ) .
the vortex - antivortex pair travels in the negative y - direction towards the impurity ( or a third vortex ) at the origin .
the scattering angle is @xmath25 ( here @xmath34 ) . ] and @xmath55 respectively .
the vortex - antivortex pair is scattered by the impurity , deflecting to the left by an angle @xmath25 .
* ( c ) : trapping scenario * for @xmath56
. the vortex ( red trajectory ) is trapped by the impurity , and the antivortex ( blue trajectory ) orbits around it . *
( d , e ) : go - around scenario * for @xmath57 and @xmath58 respectively .
the vortex and the antivortex overtake the impurity , going around it in opposite directions .
, title="fig : " ] and @xmath55 respectively .
the vortex - antivortex pair is scattered by the impurity , deflecting to the left by an angle @xmath25 . * ( c ) : trapping scenario * for @xmath56 .
the vortex ( red trajectory ) is trapped by the impurity , and the antivortex ( blue trajectory ) orbits around it . * ( d , e ) : go - around scenario * for @xmath57 and @xmath58 respectively . the vortex and the antivortex
overtake the impurity , going around it in opposite directions .
, title="fig : " ] and @xmath55 respectively .
the vortex - antivortex pair is scattered by the impurity , deflecting to the left by an angle @xmath25 . *
( c ) : trapping scenario * for @xmath56 .
the vortex ( red trajectory ) is trapped by the impurity , and the antivortex ( blue trajectory ) orbits around it . *
( d , e ) : go - around scenario * for @xmath57 and @xmath58 respectively . the vortex and the antivortex
overtake the impurity , going around it in opposite directions .
, title="fig : " ] and @xmath55 respectively .
the vortex - antivortex pair is scattered by the impurity , deflecting to the left by an angle @xmath25 .
* ( c ) : trapping scenario * for @xmath56 . the vortex ( red trajectory )
is trapped by the impurity , and the antivortex ( blue trajectory ) orbits around it .
* ( d , e ) : go - around scenario * for @xmath57 and @xmath58 respectively .
the vortex and the antivortex overtake the impurity , going around it in opposite directions .
, title="fig : " ] and @xmath55 respectively .
the vortex - antivortex pair is scattered by the impurity , deflecting to the left by an angle @xmath25 . *
( c ) : trapping scenario * for @xmath56 .
the vortex ( red trajectory ) is trapped by the impurity , and the antivortex ( blue trajectory ) orbits around it . *
( d , e ) : go - around scenario * for @xmath57 and @xmath58 respectively .
the vortex and the antivortex overtake the impurity , going around it in opposite directions .
, title="fig : " ] + and @xmath55 respectively .
the vortex - antivortex pair is scattered by the impurity , deflecting to the left by an angle @xmath25 . * ( c ) : trapping scenario * for @xmath56 .
the vortex ( red trajectory ) is trapped by the impurity , and the antivortex ( blue trajectory ) orbits around it . *
( d , e ) : go - around scenario * for @xmath57 and @xmath58 respectively . the vortex and the antivortex
overtake the impurity , going around it in opposite directions .
, title="fig : " ] and @xmath55 respectively .
the vortex - antivortex pair is scattered by the impurity , deflecting to the left by an angle @xmath25 . * ( c ) : trapping scenario * for @xmath56 .
the vortex ( red trajectory ) is trapped by the impurity , and the antivortex ( blue trajectory ) orbits around it . *
( d , e ) : go - around scenario * for @xmath57 and @xmath58 respectively . the vortex and the antivortex
overtake the impurity , going around it in opposite directions .
, title="fig : " ] and @xmath55 respectively .
the vortex - antivortex pair is scattered by the impurity , deflecting to the left by an angle @xmath25 . * ( c ) : trapping scenario * for @xmath56 .
the vortex ( red trajectory ) is trapped by the impurity , and the antivortex ( blue trajectory ) orbits around it . *
( d , e ) : go - around scenario * for @xmath57 and @xmath58 respectively . the vortex and the antivortex
overtake the impurity , going around it in opposite directions .
, title="fig : " ] and @xmath55 respectively .
the vortex - antivortex pair is scattered by the impurity , deflecting to the left by an angle @xmath25 . * ( c ) : trapping scenario * for @xmath56 .
the vortex ( red trajectory ) is trapped by the impurity , and the antivortex ( blue trajectory ) orbits around it . *
( d , e ) : go - around scenario * for @xmath57 and @xmath58 respectively . the vortex and the antivortex overtake the impurity , going around it in opposite directions .
, title="fig : " ] and @xmath55 respectively .
the vortex - antivortex pair is scattered by the impurity , deflecting to the left by an angle @xmath25 . * ( c ) : trapping scenario * for @xmath56 .
the vortex ( red trajectory ) is trapped by the impurity , and the antivortex ( blue trajectory ) orbits around it . *
( d , e ) : go - around scenario * for @xmath57 and @xmath58 respectively .
the vortex and the antivortex overtake the impurity , going around it in opposite directions .
, title="fig : " ] and @xmath55 respectively .
the vortex - antivortex pair is scattered by the impurity , deflecting to the left by an angle @xmath25 .
* ( c ) : trapping scenario * for @xmath56 . the vortex ( red trajectory )
is trapped by the impurity , and the antivortex ( blue trajectory ) orbits around it .
* ( d , e ) : go - around scenario * for @xmath57 and @xmath58 respectively .
the vortex and the antivortex overtake the impurity , going around it in opposite directions .
, title="fig : " ] showing the emission of a sound wave during trapping , corresponding to the evolution shown in fig .
[ fig2](c ) , at two different times @xmath59 ( left ) and @xmath60 ( right ) .
the red ( right ) line and dot mark the vortex and its trajectory , the blue ( left ) line and dot mark the antivortex . at @xmath61
the vortex and the antivortex are respectively at @xmath62 and @xmath63 ; at @xmath64 they are at @xmath65 and @xmath66 .
, title="fig:"][fig : sound2 ] showing the emission of a sound wave during trapping , corresponding to the evolution shown in fig .
[ fig2](c ) , at two different times @xmath59 ( left ) and @xmath60 ( right ) .
the red ( right ) line and dot mark the vortex and its trajectory , the blue ( left ) line and dot mark the antivortex . at @xmath61 the vortex and the antivortex are respectively at @xmath62 and @xmath63 ; at @xmath64 they are at @xmath65 and @xmath66 .
, title="fig:"][fig : sound3 ] as a function of impact parameter @xmath7 for initial vortex - antivortex separation @xmath38 ( top ) and @xmath37 ( bottom ) when the target is an impurity ( red line and dots ) and a vortex ( blue line and dots ) .
the shaded red areas represent the parameter regions where @xmath25 can not be defined because one vortex becomes trapped in the impurity . comparing top and bottom ,
notice the absence of the trapping regime and the extension of the figure to larger positive values of @xmath7 ( negligible deflection if the target is an impurity , non - negligible if it is a vortex ) .
, title="fig : " ] + as a function of impact parameter @xmath7 for initial vortex - antivortex separation @xmath38 ( top ) and @xmath37 ( bottom ) when the target is an impurity ( red line and dots ) and a vortex ( blue line and dots ) .
the shaded red areas represent the parameter regions where @xmath25 can not be defined because one vortex becomes trapped in the impurity . comparing top and bottom ,
notice the absence of the trapping regime and the extension of the figure to larger positive values of @xmath7 ( negligible deflection if the target is an impurity , non - negligible if it is a vortex ) .
, title="fig : " ] ( left ) and phase(right ) .
, title="fig:"][fig : large_v_d ] ( left ) and phase(right ) .
, title="fig:"][fig : large_v_p ] + ( left ) and phase(right ) .
, title="fig:"][fig : shallow_v_d ] ( left ) and phase(right ) .
, title="fig:"][fig : shallow_v_p ]
|
understanding quantum dynamics in a two - dimensional bose - einstein condensate ( bec ) relies on understanding how vortices interact with each others microscopically and with local imperfections of the potential which confines the condensate . within a system consisting of many vortices ,
the trajectory of a vortex - antivortex pair is often scattered by a third vortex , an effect previously characterised .
however , the natural question remains as to how much of this effect is due to the velocity induced by this third vortex and how much is due to the density inhomogeneity which it introduces . in this work
, we describe the various qualitative scenarios which occur when a vortex - antivortex pair interacts with a smooth density impurity whose profile is identical to that of a vortex but lacks the circulation around it .
|
graph layout problems are a class of combinatorial optimization problems whose goal is to find a layout of an input graph _
g _ to optimize a certain objective function . a linear layout or layout of an undirected graph @xmath1 , where @xmath2 is the bijective function @xmath3=\{1,2,\dots,\emph{n}\}$ ] .
set of all layouts is denoted by @xmath4 .
a region in layout is defined as the area between two consecutive vertices in the layout .
vertex separation minimization problem ( vsmp ) is to find a layout @xmath5 of a graph @xmath0 which minimizes the vertex separation ( _ vs _ ) where @xmath6}\delta(i,\varphi , g)$ ] for a layout @xmath7 where , @xmath8 , @xmath9 and @xmath10 @xcite .
vsmp is np - complete in general and has applications in vlsi design , graph drawing and computer language compiler design @xcite .
+ + further , in this paper , a vertex identifier corresponds to the vertices of graphs which are assumed to be the natural numbers 1,
,_n_. for a vertex @xmath11 , neighbourhood of _ u _ in _ g _ , @xmath12 and degree of _ u _ in _ g _ , @xmath13 .
[ fig1](b ) represents a linear layout of fig .
[ fig1](a ) .
_ vs _ for this layout @xmath7 of _ g _ is 3 .
, width=339,height=137 ] vsmp is polynomially solvable for grids , trees , permutation graphs and cographs @xcite .
this problem has been explored using metaheuristics namely , variable neighbourhood search by @xcite in which they have proposed construction heuristics to generate an initial solution .
noberto et al .
@xcite highlight the importance of construction heuristics in metaheuristics .
they have proposed eight construction heuristics and have compared all the construction heuristics . in this paper , we have proposed three construction heuristics _ _ h__1 , _ _
h__2 , _ _ h__3 .
an experiment has been performed to compare all the construction heuristics on the test instances of grids , small graphs , trees and harwell boeing graphs , totaling 248 instances of graphs .
experiments reveal that _ _
h__1 , _ _ h__2 and _
_ h__3 are able to achieve best results for 87.9% , 43.5% and 37.1% of the total instances respectively , while the best construction heuristic in the literature achieves the best solution for 39.9% of the total instances .
we have also compared the results of _ _ h__1 with the state - of - the - art metaheuristic general variable neighbourhood search ( gvns ) @xcite and observed that the proposed construction heuristics improves the results for some of the input instances .
it was also observed that gvns obtains best results for 82.9% instances of all input instances while the heuristic _ _ h__1 obtained best results for 82.3% instances of all input instances .
further , in this paper section 2 presents the construction heuristics proposed in this paper followed by computational experiments to show the efficiency of heuristics in section 3 .
in this section , three construction heuristics have been proposed . these heuristics
construct a layout by iteratively adding vertices to an initially empty layout . in each iteration , heuristic _
_ h__1 tries to place a vertex _ u _ in the partial layout which minimizes its contribution as well as the contribution of vertices which are already placed in the partial layout .
motivation behind _ _ h__2 and _ _
h__3 is that if adjacent vertices are closely placed then they contribute to fewer regions .
further , in order to minimize the number of regions to which a vertex contributes , vertices of smaller degree are placed first in the partial layout . in the following sections , we describe the three construction heuristics which are essentially greedy methods .
this procedure is outlined in algorithm 1 .
let _ layout _ be the ( partial ) layout under construction .
the set _ unvisited _ is the set of vertices of graph _
g _ which are not in _ layout_. the time complexity of _ _ h__1 is @xmath14 .
step 1 : : : @xmath15 step 2 : : : select a vertex @xmath16 such that @xmath17 is least step 3 : : : @xmath18 step 4 : : : @xmath19 step 5 : : : * while * @xmath20 step 6 : : : @xmath21 is adjacent to least number of vertices in @xmath22 } step 7 : : : @xmath23 is adjacent to largest number of vertices in @xmath24 } step 8 : : : @xmath25 is adjacent to least number of vertices in @xmath22 } step 9 : : : select @xmath26 randomly step 10 : : : @xmath27 step 11 : : : @xmath19 step 12 : : : * endwhile * algorithm 2 presents heuristic _ _
time complexity of heuristic _
_ h__2 is @xmath28 .
step 1 : : : u = minimium degree vertex of a graph g step 2 : : : @xmath29 step 3 : : : @xmath30 step 4 : : : s = sort n(u ) in ascending order of the degrees step 5 : : : @xmath31 step 6 : : : @xmath32 step 7 : : : * while * @xmath33 step 8 : : : @xmath34 step 9 : : : select a vertex v such that @xmath35 is least but non - zero step 10 : : : * if * @xmath36 step 11 : : : @xmath27 step 12 : : : @xmath37 step 13 : : : * endif * step 14 : : : @xmath38 step 15 : : : @xmath39 = sort @xmath24 according to their values in degree in ascending order step 16 : : : @xmath40 step 17 : : : @xmath41 this heuristic is similar to _ _
h__2 , but as opposed to the selection in step 9 of heuristic _ _
h__2 , preference is given to a vertex _ v _ in the partially constructed layout with least @xmath42 .
time complexity of _
_ h__3 is same as _ _ h__2 .
in this section we present the computational experiments that were performed to test the effectiveness of the proposed construction heuristics .
the code was implemented in matlab 7.0 and all the experiments were conducted on an intel i3 based system with 4 gb ram . the experiment to compare the construction heuristics
was performed on the four sets of instances previously used for this problem @xcite .
the test set includes small , harwell - boeing ( _ hb _ ) graphs , grids and trees . since gvns was tested on harwell - boeing ( _ hb _ ) graphs , grids and trees in the literature , therefore , construction heuristic _
_ h__1 and gvns are compared on these instances only .
the order of graph ranges from 9 to 2916 . for each heuristic _ _ h__1 , _
_ h__2 and _ _ h__3 , 30 runs were carried out and the best value of vertex separation was recorded . in this section
the proposed construction heuristics are compared with those available in the literature .
_ random _ represents randomly generated solution .
_ _ c__1 and _ _ c__2 are the construction heuristics proposed in @xcite and _ _ ha__1 , _ _
ha__2 , _ _
ha__3 , _ _
ha__4 , _ _
hn__1 , _ _
hn__2 , _ _ hn__3 and _ _ hn__4 have been proposed in @xcite .
heuristics _ _ h__1 , _ _ h__2 and _ _
h__3 have been presented in this paper .
table [ t1 ] presents the average vertex separation for different construction heuristics .
the average vertex separation over all the instances are given in the last column @xmath43 .
results show that _ _
h__1 outperforms all the heuristics .
table [ t2 ] presents number of best solution achieved using each heuristic . in the table
, the last column @xmath44 represents the total number of instances for which heuristic achieves the best value .
number in bold indicates that the value is best in that column .
heuristic _ _
h__1 achieves best result for 220 instances out of 248 instances which is equal to 88.71% of the total instances .
similarly , _ _ h__2 and _ _ h__3 achieve best results for 43.55% and 37.1% instances of total instances . among the existing heuristics , a maximum of 39.92% of the total instances best solutions are obtained by _
_ hn__1 .
.average vertex separation for different construction heuristics [ cols="^,^,^,^,^,^",options="header " , ]
in this paper , we have proposed three polynomial time construction heuristics and compared with other construction heuristics .
it was observed that these construction heuristics outperform other construction heuristics given in literature .
the best performing construction heuristic _ _ h__1 was also compared with the state - of - the - art metaheuristic gvns .
this construction heuristic achieves best results in larger number of cases than gvns . since
a good initial population is expected to converge faster to a near optimal / optimal solution , therefore , initial solutions generated using this construction heuristic in any metaheuristic will lead to high quality solutions .
diaz , j. , j. petit and m. serna , _ a survey of graph layout problems _ , acm computing surveys .
* 34 * ( 2002 ) , 313356 .
duarte a. , l. f. escudero , r. marti , n. mladenovic , j. j. pantrigo and j. sanchez - oro , _ variable neighborhood search for the vertex separation problem _ , computers and operations research .
* 39 * ( 2012 ) , 32473255 .
garcia n. c. , h. j. f. huacuja , j. a. m. flores , r. a. p. rangel , j. j. g. barbosa , j. m. c. valadez , _ comparative study on costructive heuristics for the vertex separation problem _ , design of intelligent systems based on fuzzy logic , neural networks and nature - inspired optimization .
( 2015 ) , 465474 .
petit , j. , _ addenda to the survey of layout problems _ , bulletin of the eatcs .
* 105 * ( 2011 ) , 177201 .
sanchez - oro j. , j. j. pantrigo and a. duarte , _ combining intensification and diversification strategies in vns .
an application to the vertex separation problem .
_ , computers and operations research . * 52 * ( 2014 ) , 209219 .
|
vertex separation minimization problem ( vsmp ) consists of finding a layout of a graph @xmath0 which minimizes the maximum vertex cut or separation of a layout .
it is an np - complete problem in general for which metaheuristic techniques can be applied to find near optimal solution .
vsmp has applications in vlsi design , graph drawing and computer language compiler design .
vsmp is polynomially solvable for grids , trees , permutation graphs and cographs .
construction heuristics play a very important role in the metaheuristic techniques as they are responsible for generating initial solutions which lead to fast convergence . in this paper , we have proposed three construction heuristics _ _ h__1 , _ _ h__2 and _ _ h__3 and performed experiments on grids , small graphs , trees and harwell boeing graphs , totaling 248 instances of graphs .
experiments reveal that _ _
h__1 , _ _ h__2 and _ _ h__3 are able to achieve best results for 88.71% , 43.5% and 37.1% of the total instances respectively while the best construction heuristic in the literature achieves the best solution for 39.9% of the total instances .
we have also compared the results with the state - of - the - art metaheuristic gvns and observed that the proposed construction heuristics improves the results for some of the input instances .
it was found that gvns obtained best results for 82.9% instances of all input instances and the heuristic _ _
h__1 obtained best results for 82.3% of all input instances .
|
the combined analysis of the type ia supernovae , galaxy clusters measurements and wmap ( wilkinson microwave anisotropy probe ) data brings out clearly an evidence of the accelerated expansion of the universe @xcite-@xcite .
the cosmological acceleration strongly indicates that the present day universe is dominated by a smoothly distributed slowly varying cosmic fluid with a negative pressure , the so - called dark energy @xcite-@xcite ( alternative approaches are presented , for example , in @xcite ) . to specify different types of cosmic fluids one
usually uses a phenomenological relation between the pressure density @xmath0 and the energy density @xmath1 , corresponding to each component of fluid @xmath2 where @xmath3 is the equation - of - state parameter or , for short , the state parameter . a component with negative
@xmath3 corresponds to the dark energy .
contemporary experiments , including wmap , give strong support that at present time the dark energy state parameter is close to @xmath4 @xcite , @xcite-@xcite .
in particular , it follows from the current observational bounds @xcite that @xmath3 belongs to the interval : @xmath5 at the @xmath6 confidence level . from the theoretical point of view
the above - mentioned domain of @xmath3 covers three essentially different cases : @xmath7 and @xmath8 . * the first case , @xmath9 , is achieved in quintessence models @xcite-@xcite , which are cosmological models with a scalar field .
such types of models are quite acceptable , but there is a question of an origin of the scalar field . to comply with astronomical experimental data
this scalar field should be extra light and hence it does not belong to the standard model set of fields @xcite . * the second case , @xmath10 ,
is realized by means of the cosmological constant .
this scenario is admissible from a general point of view except for a problem of an order of the magnitude of the cosmological constant
. it should be @xmath11 times less than the natural theoretical prediction @xcite . * the third case , @xmath8 ,
is called a `` phantom '' one and can be realized due to a scalar field with a ghost ( phantom ) kinetic term . in this case all natural energy conditions are violated and there are problems of instability at classical and quantum levels @xcite-@xcite . since experimental data do not contradict with a possibility of @xmath8 and moreover a direct search strategy to test the inequality @xmath8 has been proposed @xcite , it is interesting to find a noncontradictory model with the condition @xmath8 satisfied is a result of axion consideration . ] .
note the @xmath12-essence models @xcite can have both @xmath8 and @xmath13 . at the same time
a dynamical transition from the region @xmath13 to the region @xmath8 or vice versa is forbidden under general assumptions @xcite and is possible only under special conditions @xcite .
let us recall that in models with a constant state parameter @xmath3 , less than @xmath4 , and the spatially flat friedmann metric the scale factor tends to infinity and therefore the universe breaks down at a finite moment of time .
this problem is known as the `` big rip '' @xcite , see also @xcite . the simplest way to overcome
this difficulty in models with @xmath8 is to consider a scalar field @xmath14 with a negative time component in the kinetic term @xcite .
the appearance of a such summand is possible due to quantum effects @xcite .
however , all proposed models insist of the instability problem . a possible way to evade the instability problem for models with @xmath8 is to yield a phantom model as an effective one , arising from a more fundamental theory without a negative kinetic term . in particular , if we consider a model with higher derivatives such as @xmath15 , then in the simplest approximation : @xmath16 , such a model gives a kinetic term with a wrong ( ghost ) sign .
it turns out , that such a possibility does appear in the string field theory framework @xcite . since the concerned model is a string theory approximation , all stability problems related to a model with higher derivatives
are discarded .
our goal in the present paper is to construct an analytic solution in a polynomial model , which is close in some approximation to a model arising in the string theory , namely in the theory of a fermionic nsr string with gso@xmath17 sector .
the scalar field @xmath14 is an open string theory tachyon , which , according to the sen s conjecture @xcite , describes brane decay , at which a slow transition to the stable vacuum , correlating with states of the closed string , takes place .
other dark energy models , which use the brane - world scenarios , are presented in @xcite .
the notable feature of the model is a ghost sign of the kinetic term and a special form of the potential : @xmath18 in the first approximation only the first summand in the right hand side of ( [ vphi ] ) can be obtained from the open string field theory in a flat metric @xcite ( which corresponds to @xmath19 ) .
the second summand allows us to save the form of the interpolating solution obtained in the flat metric for the case of the friedmann universe , i.e. at an arbitrary @xmath20 .
note , that the appearance of higher powers of @xmath14 is possible if we take into account the heavier perturbations of the string @xcite , but we can not expect to obtain the explicit values of the coefficients of the potential ( [ vphi ] ) from the string theory . the performed in this paper stability analysis of solutions points out that the constructed solution is an attractive one and there are no essential changes in the behavior of the solutions for some variations of the potential parameters .
this gives reasons to assume , that the qualitative behavior of the physical parameters @xmath21 in the model with the potential ( [ vphi ] ) under certain conditions becomes close to the behavior of the corresponding ones in the model with a string theory potential .
the constructed solution has no singular points and becomes the de - sitter solution at large time .
it does not contradict to the general theorems of the relativity theory because the energy conditions are violated in the considered model ( this question is discussed in @xcite ) .
the paper is organized as follows . in section 2
we describe the cosmological model , which is an approximation for the string inspired model , and write down the solutions to the friedmann equations with the special polynomial potential . in section 3
we consider the phantom dynamics in detail and find out that our solution describes an accelerating universe . in section 4 we study the stability of the solution with respect to variations of initial data and peculiar variations of a form of the potential . in section 5 we do sum up the obtained results and point out the ways for generalization of the considered model .
let us consider the gravitational model with a phantom scalar field and the spatially flat friedmann metric .
since the origin of the phantom field , considering in this paper , is connected with the string field theory we include the typical string mass @xmath22 and a dimensionless open string coupling constant @xmath23 in the action .
the action is : @xmath24 where @xmath25 is the planck mass , @xmath26 is a spatially flat friedmann metric : @xmath27 where @xmath28 is the scale factor .
the coordinates @xmath29 and the field @xmath14 are dimensionless .
hereafter we use the dimensionless parameter @xmath30 .
independent equations of motion are as follows @xmath31 here @xmath32 is the hubble parameter : @xmath33 and the dot denotes the time derivative .
if the scalar field depends only on time , i.e. @xmath34 , then @xmath35 we recast system ( [ eomprho ] ) to the following form @xmath36 besides of equations ( [ eom12 ] ) one can obtain from the action ( [ action ] ) an equation of motion for the field @xmath14 @xmath37 where @xmath38 .
this equation is in fact a consequence of system ( [ eomprho ] ) . expressing @xmath32 through @xmath14 and @xmath39
, we obtain @xmath40 here @xmath41 factor selects the proper branch of the square root .
the expanding universe corresponds to @xmath42 . at present time one of the possible scenarios of the universe evolution
considers the universe to be a d3-brane ( 3 spatial and one time variable ) embedded in higher - dimensional space - time .
d - branes do naturally emerge in the open string theory ( in our case we consider fermionic open strings containing both gso@xmath43 and gso@xmath17 sectors ) .
the d - brane in question is unstable and does evolve to the stable state .
this process is described by the open string dynamics , which ends are attached to the brane ( see reviews @xcite and references therein ) . if only the lowest excitation the tachyon is taken into account then the d - brane dynamics are described by an action of the open string tachyon .
there are two common ways to describe the tachyon behavior : dirac - born - infeld ( dbi ) approximation and the level truncation scheme in the covariant string field theory @xcite .
the level truncation method advocated itself checking the sen s conjecture and , therefore , it is natural to use it to analyze the dynamics of the d - brane , see @xcite .
the action for the tachyon resulting from the fermionic string field theory in the approximation of the slow - varying auxiliary field @xcite has the form @xmath44 where @xmath45 is the dimensionless tachyon field and the coordinates and the constant @xmath23 are also dimensionless .
the tilde means an action of the non - local operator @xmath46 on the field @xmath45 .
the subscript `` flat '' hereafter indicates the flat background metric is used .
note that in the bosonic string case the interaction is cubic . for the space homogenous configurations
only the dependence on time @xmath47 is left and the action takes the form @xmath48 introducing the notation @xmath49 and @xmath50 , we rewrite the equations of motion obtained from the last action in the following form @xmath51 where @xmath52 . in fact
this equation is integral one since the exponent containing an infinite number of derivatives can be replaced with an integral operator @xcite in some class of fields as follows @xmath53 note , that equation ( [ flattinteom ] ) without the first summand is an equation for the @xmath0-adic string ( in our case @xmath54 ) @xcite . in the latter case the existence of a solution interpolating between vacua has been proved @xcite .
previously this solution has been numerically constructed in @xcite .
equation ( [ flattinteom ] ) was investigated in @xcite and an existence of a rolling solution , i.e. a solution interpolating between the two vacua of the potential @xmath55 was shown numerically .
it is remarkable that the late time behavior of the rolling solution can be effectively described by a lagrangian with a ghost sign of the kinetic term and the same potential .
the corresponding action has the form @xmath56 as known , the equation of motion derived from the action ( [ flatteffact ] ) : @xmath57 has an interpolating solution a kink @xmath58 system of equations ( [ eom12 ] ) with some polynomial potential @xmath59 is not integrable .
the numeric analysis of the friedmann equations with the potential ( [ flatv ] ) has been performed in the paper @xcite .
it is known ( see , for example @xcite ) that it is always possible to find the potential @xmath59 , if either @xmath60 or @xmath61 are given explicitly .
it is worth to note that our model has common features with models considered in brane - world scenarios ( see appendix ) .
following @xcite , we assume , that @xmath61 is a function of @xmath60 , called superpotential , that is @xmath62 . using equality @xmath63
, we obtain from system ( [ eom12 ] ) : @xmath64 let us find a potential , which corresponds to the field @xmath14 of type ( [ kinkphi ] ) .
this function satisfies the following equation @xmath65 hence , from ( [ eom1w ] ) it follows @xmath66 where @xmath67 is an integration constant .
once @xmath68 is known we obtain the potential @xmath59 from ( [ eom2w ] ) .
different values of @xmath67 correspond to different @xmath59 .
the requirement that @xmath59 is an even function is equivalent to the condition @xmath69 . in this case @xmath70 it can be straightforwardly verified that if @xmath60 is a solution to ( [ eom1w ] ) , ( [ eom2w ] ) with some potential @xmath59 , then @xmath71 is a solution as well .
for example , @xmath72 also is a solution to ( [ eom1w ] ) , ( [ eom2w ] ) with the potential ( [ ourv ] ) .
indeed these two solutions correspond to the one and the same function @xmath61 , but to the two different functions @xmath68 .
we have constructed the potential @xmath73 , using an explicit form of a solution .
now we consider more general problem and look for solutions , starting with the the following two requirements : @xmath60 is a smooth function with nonzero asymptotics @xmath74 and the superpotential @xmath68 is a polynomial in @xmath14 not more than the third degree . in the first condition we can put @xmath75 without loss of generality .
the second condition guarantees that the potential @xmath59 will be a polynomial in @xmath14 and its degree is not more than six .
the above conditions allow us to rewrite equation ( [ eom1w ] ) in the following form : @xmath76 from the asymptotic conditions it follows @xmath77 taking into account ( [ conditions ] ) we solve equation ( [ subs ] ) and obtain @xmath78 we do not consider another solution : @xmath79 , because it is not a smooth function .
thus , up to a scaling and a time shift we obtain a solution , which coincides with the solution in the flat metric .
it is evident that the potential obtained from ( [ eom2w ] ) coincides with ( [ ourv ] ) up to an overall factor .
it is interesting that under above - mentioned conditions on a solution and a superpotential , the sixth degree potential is a minimal possibility for existence of an interpolating solution .
one can also see , that any solution to the equations of motion in flat space with the fourth degree potential is a solution to the friedmann equations either with the sixth degree polynomial potential or with a nonpolynomial potential @xcite .
it is interesting to note that we have built an exact solution without any approximation .
one of the standard approaches to the analysis of cosmological solutions in the presence of a scalar field is the so called slow - roll approximation which technically means a neglecting of the terms @xmath80 in equation ( [ eomphi2order ] ) and @xmath81 in the expression for the energy ( [ prho ] ) .
however , from ( [ dotphi ] ) it follows that for our solution the following relations take place : @xmath82 it is evident that the slow - roll approximation in this case is valid only when @xmath14 is close to @xmath83 .
the above - mentioned approximation describes our solution at very large times and can not be used for its description in the beginning ( close to zero ) moments of time .
the hubble parameter for the constructed solution is given by @xmath84 this parameter goes asymptotically to @xmath85 when @xmath86 , or equivalently @xmath87 .
once @xmath61 is known one readily obtains the scale factor @xmath88 where @xmath89 is an arbitrary constant .
the function @xmath28 has the following asymptotic behavior at large @xmath90 : @xmath91 it follows from formulae ( [ func_at ] ) and ( [ func_a_limit ] ) that the universe expands with an acceleration .
the deceleration parameter is negative and equal to @xmath92 the corresponding plots are drawn in fig . [ ha ] ( hereafter we assume @xmath93 for all plots if something else is not specified explicitly ) . ( left ) , @xmath94 , the hubble parameter @xmath61 ( center ) and the deceleration parameter @xmath95 ( right),title="fig:",width=188 ] ( left ) , @xmath94 , the hubble parameter @xmath61 ( center ) and the deceleration parameter @xmath95 ( right),title="fig:",width=188 ] ( left ) , @xmath94 , the hubble parameter @xmath61 ( center ) and the deceleration parameter @xmath95 ( right),title="fig:",width=188 ] + the `` jerk '' parameter @xmath96 and the `` kerk '' parameter @xmath97 are presented in fig .
[ qjk ] . + substituting the obtained solution ( [ kinkphi ] ) and the potential ( [ ourv ] ) into expressions for the pressure and energy densities , we obtain @xmath98 therefore , the parameter of state @xmath3 is given by @xmath99 note that the function @xmath100 is less than @xmath4 for @xmath101 and is equal to @xmath4 for @xmath102 .
point @xmath103 corresponds to infinite future .
the plots for the pressure density @xmath104 , the energy density @xmath105 and the parameter of state @xmath3 are presented in fig .
[ prho_fig ] .
( left ) , @xmath105 ( center ) and @xmath106 ( right),title="fig:",width=188 ] ( left ) , @xmath105 ( center ) and @xmath106 ( right),title="fig:",width=188 ] ( left ) , @xmath105 ( center ) and @xmath106 ( right),title="fig:",width=188 ] thus , in the considered model of the accelerated expanding universe the phantom field describes the dark energy with the state parameter @xmath8 .
solution ( [ kinkphi ] ) to equation ( [ eomphi ] ) is a function describing an interpolation of the field @xmath14 from point @xmath4 to point @xmath83 ( and _ not _ viceversa ) with zero initial and final velocities in the potential ( [ ourv ] ) with a friction proportional to @xmath107 , which depends on coordinate and velocity .
factor @xmath41 makes the friction negative for negative @xmath14 , i.e. the particle is accelerated by it .
indeed , the expression for friction coefficient is equal to @xmath108 on the solution for both positive and negative @xmath14 .
let us discuss on the evolution of the exactly constructed solution .
a phantom field evolution is equivalent to an evolution of the ordinary field in the flipped potential .
the shape of potential @xmath73 essentially depends on value of @xmath20 , see fig .
[ vflipplot ] , where the flipped potential @xmath109 is plotted for different values of @xmath20 . for different values of @xmath20 .
the following @xmath20 correspond to the curves ( top - down direction ) : @xmath110 , @xmath111 , @xmath112 and @xmath113 in the left plot and @xmath114 and @xmath115 in the right plot .
two vertical lines @xmath116 are used as a grid.,title="fig:",width=272 ] for different values of @xmath20 . the following @xmath20 correspond to the curves ( top - down direction ) : @xmath110 , @xmath111 , @xmath112 and @xmath113 in the left plot and @xmath114 and @xmath115 in the right plot .
two vertical lines @xmath116 are used as a grid.,title="fig:",width=272 ] to analyze the extrema , let us note the following expressions @xmath117 it is evident from ( [ vdetails ] ) that extrema @xmath116 and @xmath118 are independent of @xmath20 , whereas other extrema are functions of @xmath20 .
moreover , a particular value of @xmath20 determines whether a specific extremum is a maximum or a minimum .
possible values of @xmath20 can be separated in the following domains which determine a structure of the flipped potential @xmath109 extrema . * for @xmath19 @xmath59 is a well known double well potential . * for @xmath119 the flipped potential @xmath109 , being a 6th degree potential is similar graphically to a flipped double well potential : it has a local minimum in the point @xmath118 and maxima in the points @xmath120 . * for @xmath121 point @xmath118 in the flipped potential
becomes a local maximum two more minima appear in the interval @xmath122 , developing two wells and one hill on the way of our field @xmath14 during an interpolation from point @xmath4 to point @xmath83 , see the right plot in fig .
[ vflipplot ] . in more details : * * for @xmath123 points @xmath116 are not below the point @xmath118 in the flipped potential . * * for @xmath124 points @xmath116 are below the point @xmath118 in the flipped potential .
this means that our field @xmath14 starts at the point @xmath125 with zero initial velocity , rolls down to the well and then climbs up to the hill which is above its initial location .
this is counterintuitive while the factor @xmath41 in the friction coefficient is not taken into account .
the latter means the friction is negative for negative @xmath14 .
it seems one can disprove the last statement since it is natural to inverse the time and consider the backward motion .
it is easy to resolve this contradiction because in the case of an inverse time function @xmath32 would have an opposite sign .
* for @xmath126 the flipped potential has the only extremum which is maximum in the origin and points @xmath120 are inflection points . * for @xmath127 two minima , coming from the last multiplier of the expression @xmath128 ( [ vdetails ] ) , go beyond the interval @xmath129 and become maxima .
points @xmath116 become minima of the flipped potential .
point @xmath118 remains the maximum . in this case field @xmath14
starting at point @xmath125 climbs up to the hill from the very beginning and then rolls down and stops at the point @xmath103 .
such a behavior seems incredible and becomes real due to a negative friction for negative @xmath14 . investigating the cosmological evolution of the phantom field we use only the a part of solution ( [ kinkphi ] ) starting at the time @xmath130 . at this time
the field is at the point @xmath131 and one has to supply an initial velocity @xmath132 . for large @xmath133 ( weak coupling to the gravity ) the phantom field uses an initial kinetic energy to climb up to the hill at @xmath103 . increasing a coupling to gravity , i.e. decreasing @xmath133 we lower the height of the hill at the point @xmath103 of the flipped potential . for @xmath134 heights of the hills at points @xmath118 and
@xmath103 become equal and all initial kinetic energy is spent for the work against the friction . for @xmath135 the friction becomes stronger and the particle has the kinetic energy which is exactly enough to reach the point @xmath103 which is now lower than the hill at point @xmath118 .
let us consider the behavior of the solution to system ( [ eom12 ] ) in the neighborhood of our exact solution @xmath136 with the aim to analyze its stability .
substituting @xmath137 and @xmath138 in system ( [ eom12 ] ) , we obtain in the first order of @xmath139 the following equations : @xmath140 \displaystyle\dot\phi_1&\displaystyle= & \displaystyle\frac{\left(18 - 24m_p^2 + 24(m_p^2 - 1)\tanh(t)^2 + 6\tanh(t)^4\right)\tanh(t ) } { 12m_p^2\left(1-\tanh(t)^2\right ) } \phi_1-{}\\[2.7mm]&\displaystyle-&\displaystyle \frac{\left(3-\tanh(t)^2\right)\tanh(t)}{1- \tanh(t)^2}h_1.\\ \end{array } \label{equeps}\ ] ] system ( [ equeps ] ) has the following solution : @xmath141 where @xmath142 if @xmath143 , then @xmath144 and @xmath145 are bounded for all @xmath146 and for all values of @xmath20 . in this case
the solution can be presented in the following form : @xmath147 let us consider the case @xmath148 .
it is easy to see that if @xmath149 then @xmath145 goes to infinity as @xmath86 and , therefore , our solution is not stable . at @xmath112
we obtain from ( [ phi1h1 ] ) that @xmath150 where @xmath151 .
thus , @xmath152 and @xmath153 are bounded functions . for @xmath154 , @xmath145 and @xmath144
are bounded functions as well .
thus we obtain that our solutions are stable , the first corrections are bounded , if @xmath155 .
we remind , that for @xmath156 points @xmath116 are minima , and for @xmath112 these points are inflection ones .
to study a behavior of solutions with initial conditions much distinct from the initial conditions of the exact solutions and to answer the question whether the obtained trajectory is an attractive or repulsive one , we construct phase portraits for various values of @xmath133 .
note , that @xmath61 is a real function .
it gives a restriction on the maximal value of @xmath157 , namely , for all @xmath14 the inequality that @xmath158 is hold .
for example , let @xmath159 then only solutions with @xmath160 have the physical sense .
in other words our exact solution starts from this point with the maximal possible speed .
we can not find exact solutions to system ( [ eom12 ] ) with initial conditions @xmath159 and @xmath161 , and thus we present the results of numeric calculations . in fig .
[ phasepo ] phase portraits for @xmath162 are presented , i.e. in the domain of instability of our solution .
one can see on the graphs that evolution of solutions comes to the end at points of a minimum of flip potential @xmath109 , namely at zero for @xmath163 and @xmath111 and at the point @xmath164 for @xmath115 ( see .
[ vflipplot ] ) .
@xmath165 , @xmath111 , @xmath166 + .
thick lines strokes depict solutions with the initial condition @xmath167.,title="fig:",width=566 ] + let s consider phase portraits at @xmath168 , presented in fig .
[ phasepo05 ] . in the case
@xmath112 when @xmath116 are inflection points , a solution with the initial speed @xmath169 , as well as the exact solution tends to the point @xmath103 , however coincides with it only in this point . for @xmath113 and @xmath170
we see , that the numeric solution is close to the exact solution for @xmath171 , and the less @xmath133 the less is the value of @xmath14 , at which solutions with initial speeds @xmath167 and @xmath172 get an identical speed .
@xmath173 , @xmath113 , @xmath174 + + thus both behavior of the first corrections and phase portraits show that the obtained exact solution is stable for @xmath155 .
in this subsection we investigate an influence of variation of the coefficients in the second summand of the potential ( [ ourv ] ) on the behavior of solutions to system ( [ eom12 ] ) .
consider a dynamics of system ( [ eom12 ] ) with the potential @xmath175 parameter @xmath176 is introduced so that the points @xmath116 remain the extrema of the potential .
note , that up to a constant shift and an overall factor multiplication this is the only possible modification of an even potential ( [ ourv ] ) , which leaves the points @xmath116 to be extrema . equation ( [ eomphi ] ) differs from the kink equation of motion ( [ flateom ] ) by the friction like term and an extra proportional to @xmath177 term in the potential .
parameter @xmath176 enables us to separate the effects of the friction and an extra term in the potential and to answer the question which one provides the stability of solution . to analyze stability
it is convenient to plot phase portraits for different values of parameters @xmath20 and @xmath176 and to render them in a table as it is done in figs .
[ phaseportraits1 ] and [ phaseportraits2 ] .
guided by the plotted phase portraits and an analysis performed in sections 3 and 4 , we conclude the stabilization of solutions is due to an extra term in the potential .
namely , the obtained solution is attractive one for @xmath178 irrespectively the numeric factor in the friction term .
@xmath179@xmath180 @xmath181 + and various @xmath176.,title="fig:",width=529 ] @xmath182@xmath180 @xmath183 + and various @xmath176.,title="fig:",width=479 ]
in the present paper we have performed an analysis of an exactly solvable model of accelerating expanded universe dominated by the dark energy .
the state parameter of the model in the consideration is always less than @xmath4 and tends to @xmath4 when @xmath184 .
such a behavior of the state parameter leads to an absence of the big rip singularity in our model .
this kind of singularity does exist in the model with a constant @xmath8 .
note , that these properties of the state parameter are kept after including in our model an interaction with the cold dark matter @xcite .
it is shown also in this paper that our solution is stable with respect to small fluctuations of initial data when @xmath155 and small fluctuations of the form of the potential . the simplest way to yield an exactly solvable model with @xmath3 crossing the cosmological constant barrier @xmath10
consists in an inclusion of an additional scalar field @xcite .
authors are grateful to a.e .
pukhov and i.v .
volovich for useful discussions .
the work is supported in part by rfbr grant 05 - 01 - 00758 , i.ya . and a.k .
are supported in part by intas grant 03 - 51 - 6346 and russian president s grant nsh-2052.2003.1 , the work of s.v .
is financially supported in part by russian president s grant nsh-1685.2003.2 and grant of the scientific program `` universities of russia '' 02.02.503 .
our model as whole and system ( [ eom12 ] ) in particular are in close analogy with models considered in brane - world scenarios .
let us compare system ( [ eom12 ] ) with the system of the einstein equations in the model of gravity interacting with a single scalar field in five - dimensional space - time @xcite .
the action is @xmath185 where @xmath186 is defined as follows @xmath187 assuming the scalar field depends only on @xmath188 , the independent equations of motion are @xmath189 where @xmath190 and @xmath191 .
these equations correspond to ( [ eom12 ] ) with @xmath192 . surely , such a choice of the @xmath133 coupling is not physical , but we see that the method described in @xcite to transform ( [ eom12 ] ) into the first order differential equation works in our case as well , i.e. mathematical method , which simplifies ( [ eomwolfe1 ] ) and ( [ eomwolfe2 ] ) , does not depend on a value or sign of @xmath133 .
a. riess et al . , _ type ia supernova discoveries at @xmath194 from the hubble space telescope : evidence for past deceleration and constraints on dark energy evolution _ , astrophys
. j. * 607 * ( 2004 ) 665 ; astro - ph/0402512 .
knop et al . ,
_ new constraints on @xmath195 , @xmath196 , and @xmath3 from an independent set of eleven high redshift supernovae observed with hst _
, astrophys.j .
* 598 * ( 2003 ) 102 , astro - ph/0309368 .
s. hannestad , e. mortsell , _ probing the dark side : constraints on the dark energy equation of state from cmb , large scale structure and type ia supernovae _ ,
phys . rev .
* d66 * ( 2002 ) 063508 ; astro - ph/0205096 ; c. armendariz - picon , v. mukhanov , p.j .
steinhardt , _ a dynamical solution to the problem of a small cosmological constant and late - time cosmic acceleration _ , ( 2000 ) 4438 ; astro - ph/0004134 ; c. armendariz - picon , v. mukhanov , p.j .
steinhardt , _ essentials of k - essence _ , phys . rev .
* d63 * ( 2001 ) 103510 ; astro - ph/0006373 .
arefeva , l.v .
joukovskaya , a.s .
koshelev , _ time evolution in superstring field theory on non - bps brane .
rolling tachyon and energy - momentum conservation _
, jhep * 0309 * ( 2003 ) 012 ; hep - th/0301137 .
|
an exact solution to the friedmann equations with a string inspired phantom scalar matter field is constructed and the absence of the `` big rip '' singularity is shown explicitly .
the notable features of the concerned model are a ghost sign of the kinetic term and a special polynomial form of the effective tachyon potential .
the constructed solution is stable with respect to small fluctuations of the initial conditions and special deviations of the form of the potential .
|
since the discovery of oxypnictides lafeaso@xmath2f@xmath3@xcite , bafe@xmath4as@xmath4@xcite , lifeas@xcite and fe@xmath5se@xcite , the fe pnictides high - temperature superconductors have attracted extensive attention .
there are numerous experimental studies regarding the electronic states of the new family of iron - based superconductor@xcite .
the undoped samples exhibit the antiferromagnetic transition@xcite , and show the superconducting transition with electron doping@xcite .
the band structure calculations indicate that the fermi surfaces are composed of two hole - like cylinders around @xmath6 , a three - dimensional fermi surface , and two electron - like cylinder around m for lafeaso@xcite .
this family of iron pnictides is characterized by multi fermi surfaces , and theoretical studies have been based on multi - band models with electronic interactions@xcite .
an importance of multi - band structure is obviously exhibited in recent measurements of the fe isotope effect@xcite .
the inverse isotope effect in ( ba , k)fe@xmath4as@xmath4 can be understood by the multi - band model with competing inter - band interactions@xcite .
the two - gap theory of superconductivity has a long history , and is the generalization of the bcs theory to the case with two conduction bands@xcite the objective of this paper is to study time - reversal symmetry breaking and fractional quantum - flux vortices in multi - component superconductors .
we investigate physical properties of the superconducting state with time - reversal symmetry breaking .
the @xmath7 gap equation has such a solution if the pairing interactions satisfy some conditions . to investigate vortex solutions
, we derive the ginzburg - landau functional@xcite from the microscopic theory of multi - component superconductivity .
the importance of phase dynamics has been pointed out previously@xcite .
the phase variables of the order parameters may lead to a new state as a minimum of the ginzburg - landau potential .
if the phase of the gap function takes a fractional value other than 0 or @xmath0 , a significant state , called the chiral state@xcite , appears and the time reversal symmetry is broken .
this unconventional state is induced from the josephson terms with frustrating pairing interactions .
the signs of the josephson terms play an important role in determining the ground state .
the three - band model leads to the double sine - gordon model as a model to describe the dynamics of phase variables .
this model is regarded as a generalization of the usual sine - gordon model@xcite and also has kink solutions as in the sine - gordon model .
a new feature that appears first in the double sine - gordon model is that we have fractional-@xmath0 kink solutions in the chiral .
we can define the topological charge from the topological conserved current . in the chiral region , fractional flux vortices may exist on a domain wall of the kink . the paper is organized as follows . in section
ii , we present the model hamiltonian that is considered in this paper . in section iii , we show the model that exhibits time - reversal symmetry breaking . in section iv , the ginzburg - landau functional for three - band superconductors
is derived by using the gorkov method .
each term of this functional is expressed in terms of the matrix @xmath8 where @xmath9 is the pairing interaction between the bands @xmath10 and @xmath11 . here
we investigate the magnetic properties near the upper critical field . in section v we consider the phase dynamics of the order parameters .
we investigate the ground state of the potential composed of phase variables and show the existence of the chiral region with time reversal symmetry breaking .
the double sine - gordon model appears in the three - band model and exhibits unique properties that are not contained in the conventional sine - gordon model . in section
vi we discuss the existence of fractional flux vortices and multi - vortex bound states .
we also discuss applications to multi - band superconductors such as fe pnictides .
we adopt the three - band bcs model with the attractive interactions @xmath12 where @xmath10 and @xmath11 ( = 1,2,3 ) are band indices .
@xmath13 stands for the kinetic operator .
we assume that @xmath14 .
the second term is the pairing interaction and @xmath9 are coupling constants .
the mean - field hamiltonian is @xmath15,\end{aligned}\ ] ] where the gap function in each band is defined by @xmath16 and its complex conjugate is @xmath17 we define green s functions as follows@xcite , @xmath18 @xmath19 where @xmath20 is the time - ordering operator and we use the notation @xmath21 . in terms of the green s functions ,
the gap functions satisfy the system of equations @xmath22 this yields the gap equation , @xmath23 where @xmath24 and @xmath25 is the temperature where we set boltzmann constant @xmath26 to unity .
@xmath27 is the density of states at the fermi surface . since all the bands couple with each other through mutual interactions @xmath9
, we have one critical temperature @xmath28@xcite . at the critical temperature @xmath29 , this equation reads@xcite @xmath30 for the cutoff energy @xmath31 .
@xmath32 denotes the euler constant .
here we assume the same cutoff energy in all the interactions .
the critical temperature @xmath28 is obtained as a solution to the equation @xmath33 where @xmath34 .
the system of equations in eq.([gapeq ] ) yields a set of differential equations@xcite @xmath35 here , @xmath36 is the charge of the electron , and @xmath37 is the electron velocity at the fermi surface in the @xmath38-th band .
the planck constant @xmath39 has been dropped except in front of the vector potential @xmath40 .
we set @xmath41 .
then , the equation for @xmath42 is @xmath43 \delta_j + { \rm ln}\left(\frac{2e^{\gamma}\omega_c}{\pi t}\right)\sum_{\ell(\neq j ) } a_{j\ell}\delta_{\ell}+\cdots.\ ] ] in the second term of the right - hand side we can replace @xmath25 by @xmath28 near the transition temperature . from eq.([gaplin ] ) we obtain @xmath44 where @xmath45 are elements of the inverse of the matrix @xmath46 : @xmath47 .
then the equation for @xmath42 reads @xmath48 \delta_j + \sum_{\ell(\neq j ) } a_{j\ell}\sum_{m}b_{\ell m}\delta_m+\cdots . \label{eqj}\ ] ] for example , for @xmath49 we obtain @xmath50+\cdots,\end{aligned}\ ] ] where @xmath8 is the matrix of coupling constants . to obtain the multi - band ginzburg - landau functional ,
we multiply eq.([eqj ] ) by @xmath51 and take a summation with respect to @xmath11 .
we use the gap equation @xmath52 at @xmath29
. then the energy functional density @xmath53 is @xmath54 here we set @xmath55 , and we neglect unimportant constants @xmath56 in the following .
the fourth order term is simply given by @xmath57 .
please note that out functional is relevant near @xmath28 .
the three - band superconductor exhibits a significant state with broken time - reversal symmetry , that is not shown in single - band and two - band superconductors .
this originates from the fact that the gap equation for the three - band model has complex eigen - functions in some cases .
the gap equation is @xmath58 where @xmath59 .
we assume that the matrix @xmath60 is real and symmetric .
when two eigenvalues are degenerate , we obtain two complex eigenvectors by making linear combinations .
for example , let us consider a matrix @xmath61 given by @xmath62 where @xmath63 is a positive constant and @xmath64 is also a constant satisfying @xmath65 .
@xmath63 indicates the strength of interband coupling constants .
the degenerate eigenvalue is given by @xmath66 @xmath67 and @xmath68 are euler angles in the range of @xmath69 and @xmath70 .
the eigenvectors of the matrix in eq.([ematrix ] ) that correspond to the eigenvalue @xmath71 are @xmath72 when @xmath73 , they are @xmath74 where the angle @xmath75 satisfies @xmath76 , and we assume that @xmath77 . in this way the time - reversal symmetry broken state is obtained from the gap equation . if we set @xmath73 and @xmath78 , we obtain the case with three equivalent interband pairing interactions , that is , the three equivalent bands@xcite . in this case
we have @xmath79 , and thus the phase difference between order parameters of three bands is @xmath80 .
in this section , we derive the ginzburg - landau free energy from the bcs model to investigate chirality and kinks . from the results in the section
ii , the ginzburg - landau functional for three - gap superconductors has the form @xmath81.\end{aligned}\ ] ] @xmath82 are constants and @xmath83 is the flux quantum , @xmath84 the coefficients of bilinear terms are @xmath85,\ ] ] @xmath86 for @xmath87 .
the formula for more than three bands is straightforwardly obtained in terms of @xmath88 .
the critical temperature is given by , with @xmath89 in eq.([determ ] ) , @xmath90 for the two - band case , if we assume that gap functions are constant , @xmath89 is explicitly written as @xmath91\nonumber\\ & = & \frac{g_{11}/n_2+g_{22}/n_1}{2{\rm det}g } \nonumber\\ & -&\frac{1}{{\rm det}g}\sqrt { \frac{1}{4}\left(\frac{g_{11}}{n_2}-\frac{g_{22}}{n_1 } \right)^2+\frac{g_{12}g_{21}}{n_1n_2 } } .\end{aligned}\ ] ] this is the formula first obtained by suhl et al.@xcite in the case where three bands are equivalent , we can assume simply that @xmath92 , @xmath93 and @xmath94 ( i , j=1,2,3 ) . in this case
we have @xmath95,\ ] ] @xmath96 here we investigate the magnetic properties near the upper critical field @xmath97 .
the ginzburg - landau equations are @xmath98 @xmath99 @xmath100 let us assume that the magnetic field @xmath101 is along the z axis : @xmath102 .
we adopt the vector potential @xmath103 .
if we suppose that @xmath104 , we obtain for @xmath11=1,2,3 , @xmath105 since this equation is analogous to that for the harmonic oscillator , we can set @xmath106 where @xmath107 ( @xmath108 ) are constants and @xmath109 near the critical point where @xmath110 , we can neglect the terms @xmath111 since @xmath112 are small .
this yields the linearized equation , @xmath113 to have a nontrivial solution @xmath114 , the secular equation must hold , @xmath115 @xmath28 and the upper critical field @xmath116 is determined from this third - order secular equation . in the two - band case
this equation is simply @xmath117 the @xmath97 and the coefficients @xmath118 are immediately derived from this as @xmath119,\ ] ] @xmath120.\ ] ] let us consider a simple case for the three - band model , where we assume that three bands are equivalent : @xmath121 , @xmath122 , @xmath123 , @xmath124 , @xmath125 ( @xmath126 ) and @xmath127 .
we easily obtain @xmath128\ ] ] @xmath129 in the @xmath130-band model , these formulae are generalized to @xmath131,\ ] ] @xmath132 hence , both the upper critical field @xmath97 and the critical temperature @xmath28 increase as the number of bands is increased .
the increase of the upper critical field , that is , the reduction of the coherence length @xmath133 , due to the multi - band effect should be noticed .
in fact , the high @xmath97 has been reported for the fe pnictide smfeaso@xmath134f@xmath135@xcite .
we can show that the three - band system is governed by a single ginzburg - landau parameter @xmath136 near @xmath97 .
thus the conventional three - band model does not yields a type-1.5 superconductor@xcite .
we define @xmath137 where @xmath138 , @xmath139 is the effective mass and @xmath140 is the average of the coefficients @xmath82 ( @xmath108 )
. then we can show , following the standard procedure@xcite , that the free energy is given as @xmath141,\ ] ] where @xmath142 is the integral of @xmath143 over the space and @xmath144 is a constant .
this is the same formula as for the single - band superconductor if we substitute @xmath145 to @xmath136 .
the magnetization is similarly given by @xmath146 the multi - band effect that is contained through the parameter @xmath136 . hence , if there are no some reasons concerning the symmetry of crystals and superconductivity , the superconductor is regarded as the type 1 or type 2 .
in section iii , we have shown the model that has a solution with time - reversal symmetry breaking .
this section is devoted to an investigation of this solution in three - band superconductors by using the ginzburg - landau theory .
we now consider the phase dynamics of the order parameters .
we write the order parameters as @xmath147 where @xmath148 is a real quantity . for simplicity , we assume that the coefficients of the josephson terms are real : @xmath149 . the free energy density is denoted as @xmath53 , that is , the free energy is given by the integral of @xmath53 over the space .
@xmath53 is written as @xmath150 we focus on the role of phases of the order parameters ; we assume that @xmath151 and define new phase variables @xmath152 the free energy density is @xmath153 \nonumber\\ & - & 3k\rho\nabla^2\rho+3k\frac{4\pi^2}{\phi_0 ^ 2}\rho^2{\bf a}^2 + \frac{1}{3}k\rho^2(\nabla\phi)^2 \nonumber\\ & + & k\frac{4\pi}{\phi_0}\rho^2{\bf a}\cdot\nabla\phi + \frac{1}{8\pi}{\bf h}^2\nonumber\\ & + & \frac{1}{3}k\rho^2[(\nabla\varphi_1)^2+(\nabla\varphi_2)^2 + ( \nabla(\varphi_1+\varphi_2))^2],\end{aligned}\ ] ] where @xmath154 and @xmath155 .
the stationary conditions with respect to the fields @xmath40 , @xmath156 and @xmath157 lead to @xmath158 @xmath159 - 6k\nabla^2\rho+6k\frac{4\pi^2}{\phi_0 ^ 2}h^2x^2\rho + \frac{2}{3}k\rho(\nabla\phi)^2\nonumber\\ & + & \frac{2}{3}k\rho[(\nabla\varphi_1)^2+(\nabla\varphi_2)^2 + ( \nabla(\varphi_2+\varphi_2))^2]=0,\end{aligned}\ ] ] @xmath160=0,\ ] ] @xmath161=0.\ ] ] in this section we examine the ground state of the system with the potential @xmath162 if we set @xmath163 , we have @xmath164 ( mod@xmath165 ) .
the minimum of this potential is dependent on the signs of the coefficients @xmath166 of the josephson terms .
we define @xmath167 , @xmath168 , and @xmath169 .
the potential is written as @xmath170 we assume that the absolute values @xmath171 are almost equal in magnitude
. then there are four cases to be examined as shown in table i. when all the @xmath172 are negative , we have the minimum at @xmath173 ( case i ) .
if we change the sign of @xmath174 , this produces a frustration effect and @xmath175 take fractional values .
for example , when all the @xmath171 are equal , we have a minimum at ( @xmath176,@xmath177,@xmath178)=(@xmath179,@xmath179,@xmath180 ) . in this state
the order parameters are complex and thus the time reversal symmetry is broken .
the case iv also exhibits a similar state with fractional values of @xmath175 .
if all the @xmath171 are the same , the ground state is at ( @xmath176,@xmath177,@xmath178)=(@xmath80,@xmath80,@xmath80 ) ( fig.1 ) . for @xmath181 .
black and white dots indicate minima of the potential @xmath182 .
dotted line is the path in the valley connecting two minima .
, width=377 ] [ potential ] .classification of the ground state of the potential @xmath182 of the josephson interactions .
@xmath175 in the cases iii and iv are for @xmath183 .
[ cols="^,^,^,^,^,^,^,^",options="header " , ]
in general , there are solutions of vortices with fractional quantum flux in multi - band superconductors .
kinks in the space of phase variables @xmath184 play a central role for the existence of fractional flux vortices . in the two - band model ,
the half - quantum - flux vortex exists with a line of singularity of the phase variables @xmath184 as shown in fig.5 . here
@xmath185 changes from 0 to @xmath0 ( or @xmath0 to 0 ) across the cut , and simultaneously @xmath186 changes from 0 to @xmath187 ( or @xmath187 to 0 ) . in the case of fig.5 ,
a net - change of @xmath185 is @xmath165 by a counterclockwise encirclement of the vortex , and that of @xmath186 vanishes , due to singularities .
thus , we have a half - quantum flux vortex .
this is an interpretation of half - quantum flux vortices in triplet superconductors@xcite in terms of the phase of the order parameter .
and @xmath186 have line singularities , as shown in ( a ) and ( b ) .
two half - flux vortices connected by the singularity are shown in ( c ) .
, width=377 ] [ 2-vortex ] in three - band superconductors , the fractional - flux vortex exists in the chiral case as well as the non - chiral case .
since we have the fractional-@xmath0 kink in the chiral state ( cases iii and iv ) , the new types of vortices with fractional flux quanta exist on a domain wall of the kink@xcite .
the kink considered in the previous section is a one - dimensional structure in superconductors .
there are many types of kinks connecting two minima of the potential in three - band superconductors . .
the vortex is at the point @xmath188 with flux @xmath189 where @xmath83 is the flux quantum .
we start from @xmath190 to reach @xmath191 ( modulo @xmath165 ) through the @xmath180-kink and @xmath80-antikink .
@xmath192 goes from @xmath193 to @xmath80 crossing the @xmath180-kink , and @xmath176 goes from @xmath80 to @xmath194 ( mod @xmath165 ) through the @xmath80-kink .
, width=377 ] [ fvortex ] ( @xmath195 ) have singularities that are described by kinks in fig.6 .
the total flux is @xmath83 .
topologically , the flux @xmath196 is equivalent to @xmath197 .
thus , this state corresponds to the meson under the duality transformation between charge and magnetic flux .
, width=302 ] [ 2-fvortex ] let us discuss the fractional vortices in the three - band model here .
suppose that two kinks , one is a kink and the other is an antikink , intersect at a point @xmath188 as shown in fig.6 in a two - dimensional @xmath198-plane . if a vortex exists along the @xmath199 axis just at the point @xmath188 , the vortex should have a fractional flux quantum so that the phase change around the point @xmath188 is @xmath165 .
for @xmath200 and @xmath201 , this is shown schematically in fig.6 .
we set the phases of the order parameters @xmath190 in some region . after crossing the @xmath180 kink ,
they become @xmath202 where the phase variables @xmath176 and @xmath177 change from @xmath193 to @xmath80 . if there is also a domain wall of an antikink that starts from the point @xmath188 as in fig.6 , we have the phases @xmath191 after we cross the antikink . here
, we obtain the phase difference between the initial and final states ( see fig.6 ) . in this case
, the vortex that is located through the point @xmath188 along the @xmath199 axis should have a fractional flux quantum @xmath189 .
thus , in the chiral region of three - band superconductors , the existence of fractional vortices is easily concluded in this way .
and the total flux is @xmath83 .
the phase variables @xmath203 ( @xmath195 ) have singularities that are fractional-@xmath0 kinks .
, width=377 ] [ 3-vortex1 ] in the three - band model , the fractional vortex has two line singularities ( kinks ) in the phases of the gap function as shown in fig.6 . from fig.6
, we have a two - vortex bound state as presented in fig.7 in the chiral state .
two vortices form a molecule by two kinks .
this state may have lower energy than the vortex state with quantum flux @xmath83 since the magnetic energy @xmath204 is smaller than @xmath205 of the unit flux .
the energy of kinks is proportional to the distance @xmath206 between two fractional vortices if @xmath206 is large .
thus , the attractive interaction works between them if @xmath206 is sufficiently large .
three - vortex bound states are also formulated : they are shown in figs.8 , 9 and 10 .
the first two figures indicate bound states in the time - reversal symmetry broken state .
the last one is for the unbroken state@xcite .
these states correspond to baryons if we regard the magnetic flux as charge .
[ 3-vortex2 ] and the total flux is @xmath83 . in this state , the region including the point o has higher energy .
, width=377 ] [ 3-vortex3 ]
in this section , we briefly consider some applications . in the one - band model ,
the symmetry of cooper pairs is classified into irreducible representations and each representation is one dimensional in most cases except the two - dimensional representation for @xmath207-wave symmetry@xcite . in the one - dimensional representation ,
the phase of the order parameter is not important , except the case where two states in two different representations accidentally are degenerate . in multi - band systems ,
the relative phase between bands becomes important and plays an essential role .
a promising candidate of a multi - band superconductor is a superconductor - insulator - superconductor ( sis ) junction .
an sis tri - junction will be a three - band superconductor , interacting through weak josephson couplings , with three equivalent bands .
we hope that sis junctions or superlattice systems will be realized as artificial multi - band systems .
the fe pnictides in general contain several bands and it is probable that the order parameters in these bands belong to the same irreducible representation .
we can expect high @xmath97 due to the multiplicity of bands in fe pnictides . if we assume that the pairing interactions are frustrating ,
that is , in the cases iii and iv , or @xmath200 in the potential @xmath182 , there is a possibility that chiral superconductivity will emerge in fe pnictides superconductors .
let us examine the pairing interactions that originate from the spin susceptibility @xmath208 .
it has been shown that @xmath208 has peaks at @xmath209 and @xmath210 in multi - band models for fe pnictides@xcite . in some fe pnictides @xmath208
has a peak at @xmath211 as well as at @xmath209@xcite . in this case
we have a frustration between the pairing interaction due to @xmath212 and @xmath213 .
this may lead to a chiral @xmath214-state @xmath215 rather than @xmath216 and @xmath217 as an intermediate state of these two states .
in this paper we have investigated multi - band superconductors .
we have examined time - reversal symmetry breaking and fractional flux vortices in the three - band model .
our discussion is based on the mechanism of time - reversal symmetry breaking due to the degeneracy of the gap equation .
the other mechanism should also be investigated in future studies@xcite .
we derived the ginzburg - landau free energy for multi - component superconductors with more than two components .
the coefficient of each term is expressed in terms of the inverse of the matrix of pairing interactions @xmath8 .
we have shown that the system is determined by the single ginzburg - landau parameter @xmath136 near the upper critical field @xmath97 .
for the frustrating pairing interactions , the chiral superconducting state exists with broken time reversal symmetry . the double sine - gordon model appears as an effective theory to describe low - excitation energy states for three - component superconductors . in the chiral case
, this model has solutions of fractional-@xmath0 kinks and satisfies the inequality of bogomolnyi type with the topological charge @xmath1 .
we also discussed the existence of fractional flux vortices and that of the multi - vortex bound state in multi - band superconductors .
the kinks form domains , which we call the chiral domains , in multi - band superconductors ; this is analogous to the domains in triplet superconductors@xcite . in the chiral region of three - band superconductors , there exist several types of kink solitons .
vortices with fractional flux quanta exist on domain walls .
it is important to investigate the effect of domain walls on the vortex dynamics .
domain walls play an important role in strong pinning of vortices and will be closely related to flux - flow noise .
the generation of chiral domains by quenching superconductors is also an attractive subject .
this is a phenomenon caused by the kibble mechanism in superconductors@xcite . in experiments , the number of domains may be controlled by quenching speed and sample morphology that pins domain walls .
y. kamihara , t. watanabe , m. hirano and h. hosono , j. ame .
130 ( 2000 ) 3296 .
m. rotter , m. tegel , and d. johrendt , phys .
( 2008 ) 106006 .
m. j. pitcher , d. r. parker , p. adamson , s. j. c. herkelrath , a. t. boothroyd , s. j. clarke , chem .
commun . 5918 ( 2008 ) .
j. h. tapp , z. tang , b. lv , k. sasmal , b. lorenz , p. c. w. chu , a. m. guloy , phys .
b78 ( 2008 ) 060505 .
f. c. hsu , j .- y .
luo , k .- w .
yeh , t .- k .
chen , t .- w .
m. wu , y .- c .
lee , y .- l .
huang , y .- y .
yan , m .- k .
wu , proc .
( usa ) 105 ( 2008 ) 14262 .
y. mizuguchi , f. tomioka , s. tsuda , t. yamaguchi , y. takano , appl .
93 ( 2008 ) 14262
. g. f. chen , z. li , d. wu , g. li , w. z. hu , j. dong , p. zheng , j. l. luo , and n. l. wang , phys .
( 2008 ) 247002 .
x. h. chen , t. wu , g. wu , r. h. liu , h. chen and d. f. fang , nature 453 ( 2008 ) 761 . c. cruz , j. q. huang , j. w. lynn , j. li , w. ratcliff ii , j. l. zarestky , h. a. mook , g. f. chen , j. l. luo , n. l. wang , p. dai , nature 453 ( 2008 ) 899 .
y. nakai , k. ishida , y. kamihara , m. hirano , and h. hosono , j. phys .
77 ( 2008 ) 073701 .
l. shan , y. wang , x. zhu , g. mu , l. fang , and h .- h .
wen , europhys .
83 ( 2008 ) 57004 .
h. luetkens , h .- h .
klauss , r. khasanov , a. amato , r. klingeler , i. hellmann , n. leps , a. kondrat , c. hess , a. khler , g. behr , j. wener , and b. bchner , phys . rev .
( 2008 ) 097009 .
d. j. singh and m. h. du , phys .
( 2008 ) 237003 .
k. kuroki , s. onari , r. arita , h. usui , y. tanaka , h. kontani , and h. aoki , phys .
( 2008 ) 087004 .
h. ikeda , j. phys .
77 ( 2008 ) 123707 .
y. yanagi , y. yamakawa and y. ono , j. phys .
77 ( 2008 ) 123801 .
r. h. liu , t. wu , g. wu , h. chen , x. f. wang , y. l. xie , j. j. yin , y. j. yan , q. j. li , b. c. shi , w. s. chu , z. y. wu , x. h. chen , nature 459 ( 2009 ) 64 .
m. shirage , k. kihou , k. miyazawa , c .- h .
lee , h. kito , h. eisaki , t. yanagisawa , y. tanaka a. iyo , phys .
103 ( 2009 ) 257003 .
t. yanagisawa , k. odagiri , i. hase , k. yamaji , p. m. shirage , y. tanaka , a. iyo and h. eisaki , j. phys .
78 ( 2009 ) 094718 .
h. suhl , b. t. matthias , and l. r. walker , phys .
lett . 3 ( 1959 )
552 . j. kondo , prog .
theor . phys . 29
( 1963 ) 1 .
a. j. leggett , prog .
36 ( 1966 ) 901 .
b. t. geilikman , r. o. zaitsev , and v. z. kresin , soviet phys.- solid state 9 ( 1967 ) 642 .
a. gurevich , phys .
b67 ( 2003 ) 184515 .
m. e. zhitomirsky and v .- h .
dao , phys .
b69 ( 2004 ) 054508 .
a. izyumov and v. m. laptev , phase transitions 20 ( 1990 ) 95 .
y. tanaka , j. phys .
70 ( 2001 ) 2844
. y. tanaka , phys .
( 2002 ) 017002 .
e. babaev , phys . rev
. lett . 89
( 2002 ) 067001 .
a. j. leggett , rev .
76 ( 2004 ) 999 . y. tanaka and t. yanagisawa , j. phys .
79 ( 2010 ) 114706 .
y. tanaka and t. yanagisawa , solid state commnu .
150 ( 2010 ) 1980 .
r. rajaraman , _ solitons and instantons _
( north holland , amsterdam , 1987 ) .
l. p. gorkov , soviet phys .
jetp 9 ( 1959 ) 1364 .
v. stanev and z. tesanovic , phys .
rev . b81 ( 2010 )
134522 . c. senatore , r. flkiger , m. cantoni , g. wu , r. h. liu , and x. h. chen , phys .
b78 ( 2008 ) 054514 .
v. moshchalkov , m. menghini , t. nishio , q. h. chen , a. v. silhanek , v. h. dao , l. f. chibotaru , n. d. zhigadlo , and j. karpinski , phys .
( 2009 ) 117001 .
p. g. de gennes , _ superconductivity of metals and alloys _
, revised ed . , westview press , usa ( 1999 ) .
m. tinkham , _ introductio to superconductivity _ , 2nd ed .
, dover publications ( 2004 ) .
kee , y. b. kim and k. maki , phys .
b62 ( 2000 ) 9275 .
m. nitta , m. eto , t. fujimori and k. ohashi , arxiv:1011.2552 ( 2010 ) .
r. hlubina , phys .
b59 ( 1999 ) 9600 .
j. kondo , j. phys .
70 ( 2001 ) 808
. t. yanagisawa , new j. phys .
10 ( 2008 ) 023014 .
i. i. mazin , d. j. singh , m. d. johannes , and n. h. du , phys .
( 2008 ) 057003 .
y. imry , j. phys .
c8 ( 1975 ) 567 . m. sigrist and d. f. agterberg , prog .
102 ( 1999 ) 965 . t. w. b. kibble , j. phys .
a9 ( 1976 ) 1387 .
w. h. zurek , nature 317 ( 1985 ) 505 .
|
we investigate some significant properties of multi - band superconductors .
they are time - reversal symmetry breaking , chirality and fractional quantum flux vortices in three - band superconductors .
the bcs ( bardeen - cooper - schrieffer ) gap equation has a solution with time - reversal symmetry breaking in some cases .
we derive the ginzburg - landau free energy from the bcs microscopic theory .
the frustrating pairing interaction among fermi surfaces leads to a state with broken time - reversal symmetry , that is , a chiral solution .
the ginzburg - landau equation for three - component superconductors leads to a double sine - gordon model .
a kink solution exists to this equation as in the conventional sine - gordon model . in the chiral region of the double sine - gordon model ,
an inequality of bogomolnyi type holds , and fractional-@xmath0 kink solutions exist with the topological charge @xmath1 .
this yields multi - vortex bound states in three - band superconductors .
|
the kinematics of an off - axis neutrino beam in the approximation where the parent meson beam is divergenceless was posed as a problem to the student in these archives a long time ago .
however , the author of @xcite seems not to have noticed that the problem is amenable to an exact treatment which reveals interesting aspects hidden by the approximations he uses . in this note , we show how to implement exactly the lorentz transformation in the differential ( angle , energy ) distribution and we exhibit some features of the results .
+ among others , it is shown that a sizeable part of the flux at a given lab angle comes from neutrinos going backwards in the center of mass frame of the decaying meson .
given the approximations made in @xcite , it is not clear that they are taken into account in the results presented .
how this might have a bearing on the calculations of neutrino fluxes being performed with more realistic divergent meson beams in the preparation of upcoming experiments is unknown to the present author .
starred variables refer to the c.o.m . frame ( com ) of the decaying meson .
+ the neutrino energy in the lab is @xmath0 and its angle with respect to the beam axis is @xmath1 in the lab and @xmath2 in the decaying meson frame . for illustrative purposes
, we use a @xmath3 proton beam energy corresponding to the t2k experiment and a pion spectrum of the form @xmath4 as done in @xcite
@xmath5 + or + @xmath6 + with @xmath7 and + @xmath8 + in the second and third cases , @xmath9 which is henceforth the neutrino com energy is not fixed by the conservation laws , but it makes little difference in what follows , except that the lab energy is not rigidly related to the lab angle and meson velocity . + also note that for @xmath10 decays , the divergenceless hypothesis does not hold since @xmath10 s are themselves produced in meson decays . the forthcoming analysis should therefore be amended in this case .
+ for the two - body case , we take @xmath11 as our paradygm : + @xmath12 and @xmath13 together with @xmath14 imply that + @xmath15 and @xmath16 +
we take the decaying meson direction as polar axis in the following , and consider the meson beam as divergenceless .
+ consequently we shall not make use of the azimuthal degree of freedom and reduce the problem to two dimensions .
+ the neutrino energy - momentum vector in com is therefore @xmath17 + let @xmath18 be the @xmath19 lorentz factor and @xmath20 its velocity for @xmath21 + the lab neutrino energy - momentum vector is given by : + @xmath22 we set @xmath23 since our neutrinos are evidently considered as massless here .
this is not an approximation when only these kinematics are considered . to see , it is enough to remember that the neutrino lorentz @xmath24 in the @xmath19 rest frame is @xmath25 for a @xmath26 neutrino , whilst the @xmath24 of a @xmath27 @xmath19 is @xmath28 from this follows : @xmath29 all these equations are inversed by the substitution @xmath30 and @xmath31 + the relation for @xmath32 following from ( [ ct ] and [ st ] ) has no interest .
however , a complete picture of the relation @xmath31 is obtained by using @xmath33 which yields : @xmath34 which shows that one angle is a monotonous function of the other on the complete @xmath35 $ ] interval , notwithstanding the well known headlight effect . + the following useful relations
should also be noted : @xmath36
from equation ( [ st ] ) one sees that @xmath37 and therefore @xmath38 which shows that a given neutrino energy @xmath0 can only be found up to a maximum lab angle @xmath39 which is small for most neutrinos .
+ conversely , at a given angle @xmath1 from the ( supposedly divergenceless ) beam , the maximum neutrino energy is : @xmath40 this bound is valid , of course , for not too small angles . using ( [ en2 ] )
, we can retrieve it more accurately by calculating @xmath41 at fixed @xmath42 to get : @xmath43 it follows that the maximum lab energy at fixed @xmath42 is obtained for @xmath44 .
this correspond to @xmath45 and we find , using again ( [ en2 ] ) , that @xmath46 in conformity with ( [ emx ] ) .
these relations are illustrated on fig .
[ fig : enufpi ] so , although the bound seems to be independant of the beam , it makes sense only if there exist pions with @xmath47 and ( [ emx ] ) must be complemented by the restriction @xmath48 + the stationnarity of @xmath49 will result in many neutrinos emitted by pions around that value of @xmath24 to pile up at the same lab energy @xmath50 .
since they come from pions with @xmath24 s both below and above @xmath51 , they are emitted both forwards and backwards in the pion frame .
we will find the distributions of both components presently .
+ for larger pion energies , neutrinos at angle @xmath1 with respect to the axis are emitted more and more backwards in com with the result that their lab energy slightly decreases with @xmath52 ( see fig . [
fig : enufpi ] ) + in an analogous manner , we can retrieve the bound ( [ thm ] ) by using ( [ en2 ] ) to calculate the derivative @xmath53 at fixed @xmath0 .
we find : @xmath54 and the minimum @xmath42 at fixed lab energy is @xmath55 so that again , the stationnarity wrt @xmath24 will result in a piling up around @xmath42 of neutrinos coming from pions with energies around @xmath56 .
this is illustrated by the distributions ( derived below ) plotted on fig .
[ fig : cdist ]
in this subsection , we consider that the decaying meson velocity is fixed . therefore there is but one final state variable for a 2-body decay ( not counting the azimuth ) and the lab angular distribution is completely determined even in the case of 3-body decay of the parent meson . by differentiating ( [ inv ] ) with respect to @xmath42 , we find @xmath57 since the meson is spinless it decays isotropically in com and the @xmath58 distribution is flat on the interval @xmath59 $ ] the lab distribution of the cosine for a given meson energy is thus : @xmath60 this distribution is valid without change for 3-body decays . equations ( [ en ] ) and ( [ ct ] ) allow to calculate all the necessary partial derivatives : + @xmath61 from which we get the jacobian : @xmath62 so assuming some energy distribution @xmath63 for @xmath9 ( independant of the angles by isotropy ) we get : @xmath64 for 2-body meson decays , there is only one variable ( @xmath65 is a dirac @xmath66 ) and this reduces to the angular distribution given in ( [ dbl ] ) after elimination of @xmath66 ; the result can alternatively be expressed as a distribution in energy .
+ for 3-body meson decays , @xmath65 must be found from the decay dynamics . from ( [ en ] ) and ( [ mag ] ) @xmath50 for a given angle
is given by @xmath67 with + @xmath68 obtained when the mass of the system recoiling against the neutrino in com is minimal .
we no longer consider that the energy of the parent meson is fixed .
we assume an energy distribution , say @xmath69 for the @xmath19 mesons in the lab .
note that we use @xmath19 for definiteness .
when it comes to 3-body decays , @xmath70 would be more appropriate .
+ even assuming constant @xmath9 ( 2-body decay ) , the two final state variables @xmath0 and @xmath42 can now vary separately . to derive their distribution , we need the jacobian of the transformation @xmath71 + when performing this transformation , @xmath9 must be considered as a fixed parameter . in the case of 3-body
decays , the @xmath73 distribution found below will be simply multiplied by the @xmath9 com distribution and integrated over this variable .
+ it is much simpler to calculate the partial derivatives of the final variables with respect to @xmath52 and @xmath58 than the contrary .
all we need is again contained in the two relations ( [ en ] ) and ( [ ct ] ) and the relation @xmath74 the necessary derivatives are as follows : @xmath75 therefore we get : @xmath76 using the relations established above , the inverse jacobian reduces to : @xmath77 which , not surprisingly , is singular when @xmath55 , corresponding to @xmath78 + to get the final @xmath79 distribution for given @xmath9 , we must invert the relation yielding @xmath80 as a function of @xmath81 . this
is done as follows : + from ( [ ct ] ) and ( [ st ] ) one easily deduce the following equations : @xmath82 we further define : @xmath83 so that ( [ st ] ) reads @xmath84 + hence ( [ gm1 ] ) and ( [ be1 ] ) become : @xmath85 with @xmath86 according to whether @xmath87 or @xmath88 , that is , @xmath89 or @xmath90 ( cf .
( [ inv ] ) ) ; observe that ( [ be1 ] ) implies that @xmath58 is always smaller than @xmath42 .
other constraints on @xmath91 and @xmath24 are automatically obeyed by the expressions ( [ gm1 ] ) and ( [ be1 ] ) + we also get from ( [ be2 ] ) @xmath92 this allows us to rewrite the jacobian in terms of laboratory neutrino variables as : @xmath93 to see when each of the two _ a priori _ possible values of @xmath58 apply , we observe first that ( [ en ] ) and ( [ en2 ] ) show that the physical range of @xmath94 is @xmath95\ ] ] further , these two equations allow us to write @xmath42 and @xmath58 as functions of @xmath94 and the boost parameters as follows : @xmath96 the expression for @xmath58 solves the @xmath97 problem : * for @xmath98 the two signs of @xmath58 are possible since we can ( if the beam permits ! ) have @xmath99 and @xmath100 the first case corresponds to a lower energy meson with the neutrino going forward in its rest frame , whilst the second means a more energetic meson needed to compensate the backward projection of the neutrino in this frame . * for @xmath101
there is but one possibility left , corresponding to + @xmath102 * for @xmath103 , only the @xmath104 solution exists . when @xmath105 , @xmath106 * @xmath107 corresponds to neutrinos going backwards in the lab .
this is where we diverge from the treatment of @xcite .
equation @xmath108 of this author seems to imply that only the @xmath109 contribution is taken into account .
however , @xmath110 corresponds to @xmath111 which is generally very small , so that a sizeable number of almost forward neutrinos can be forgotten here .
[ fig : endist1 ] strikingly illustrates this remark .
+ therefore , to get the joint distribution for @xmath112 we must add the contributions of these two solutions when both apply .
the joint distribution for @xmath52 and @xmath113 which is simply @xmath114 is transformed to : @xmath115 \left [ g(m_{\pi}\gamma_+)|{\cos{\theta}}-\sqrt{1-r^2{\sin{\theta}}^2}|+g(m_{\pi}\gamma_-)|{\cos{\theta}}+\sqrt{1-r^2{\sin{\theta}}^2}|\right ] \end{gathered}\ ] ] where @xmath116 means ( [ gm2 ] ) with @xmath86 and _ only the second term is to be kept if @xmath117 _ .
+ this must be multiplied by the probability density of @xmath9 and integrated over it to get the final distribution in ( @xmath118 ) . in the case of a 2-body decay , this results in the mere replacement @xmath119 in ( [ fin ] ) , where the value of @xmath120 stems from the conservation laws .
figures [ fig : endist1 ] , [ fig : endist2 ] and [ fig : cdist ] show the behaviour expected for constant @xmath0 or constant @xmath42 .
the last displays jacobian peaks at @xmath121 corresponding to the limits explicited in ( [ thm ] ) .
the first shows the same phenomenon for fixed @xmath1 with piling up at @xmath122 in conformity with ( [ emx ] ) the effect seems to disappear ( fig [ fig : endist2 ] ) for @xmath123 .
as explained after equation ( [ emx ] ) , the limit ( in @xmath0 ) and the concomitant jacobian peak exists only if pions of energy larger than @xmath56 are present in the beam . with @xmath124 and the spectrum shape adopted here ,
this corresponds to @xmath125 .
however , the pion spectrum decreases sharply before this limit .
+ note that although the contribution of neutrinos going backwards in the pion rest frame dies off when going to smaller angles due to the scarcity of energetic pions , they concentrate at @xmath126 , reinforcing the peak which exists already in the distribution of neutrinos going forward in this frame .
equation ( [ fin ] ) is the final result of this paper .
it shows that there exist in general ( for @xmath127 ) two contributions to the spectrum of neutrinos emitted at a given laboratory angle , provided the meson beam is energetic enough to bring the neutrino emitted backwards in the meson frame to the required forward angle . in fig .
[ fig : endist1 ] this second contribution is displayed explicitely .
|
we review the kinematics of a neutrino beam in the idealized case where the parent mesons momenta are parallel , but without any other approximation .
this reveals several interesting features , in particular in the off - axis case , which are hidden by the approximations made in a previous treatment .
|
type ia supernovae ( sne ia ; see filippenko 1997 for a review of supernova classification ) are the thermonuclear runaway explosions of carbon / oxygen white dwarfs ( see , e.g. , hillebrandt & niemeyer 2000 for a review ) .
they can be used as standardizable candles with many important applications , including measurements of the expansion rate of the universe ( riess et al .
1998 ; perlmutter et al . 1999 ) .
two general scenarios are favored as the progenitor system for sne ia .
one is the single - degenrate model ( hoyle & fowler 1960 ; hachisu et al .
1996 ; meng et al . 2009 ; rpke et al . 2012 ) , which consists of a single white dwarf accreting material from a companion . the other is the double - degenerate scenario involving the merger of two white dwarfs ( webbink 1984 ; iben & tutukov 1984 ; pakmor et al . 2012 ; rpke et al . 2012 ) .
however , our understanding of their progenitor systems and explosion mechanisms remains substantially incomplete .
very early discovery and detailed follow - up observations are essential for understanding those problems .
for example , bloom et al .
( 2012 ) were able to constrain the companion - star radius to be @xmath9 from an optical nondetection just 4hr after the explosion of sn 2011fe ( nugent et al .
cao et al . ( 2015 ) found strong but declining ultraviolet emission in sn ia iptf14atg in early - time _
observations , consistent with theoretical expectations of the collision between supernova ( sn ) ejecta and a companion star ( kasen 2010 ) .
i m et al .
( 2015 ) found evidence of a dark phase " in sn 2015f , which can last for a few hours to days between the moment of explosion and the first observed light ( e.g. , rabinak , livne , & waxman 2012 ; piro & nakar 2013 , 2014 ) ; see also cao et al .
( 2016 ) for the case of iptf14pdk .
differences in the duration of the dark phase " could be caused by a varying distribution of @xmath10ni near the surface of a sn ia .
for example , piro & morozova ( 2016 ) show that it is short ( with a steep rise ) when the @xmath10ni is shallow , and longer ( with a more gradual rise ) when the @xmath10ni is deeper ( see their figure 7 ) .
spectra of sne ia not only reveal the ejecta composition from nuclear burning , but also provide a way to measure the ejecta expansion velocity .
benetti et al .
( 2005 ) separated sn ia samples into different groups according to their velocity gradient and found that high - velocity - gradient objects tend to have a higher velocity of the @xmath36355 line near maximum light .
nugent et al .
( 1995 ) quantified the spectral diversity using line - strength ratios , finding a good correlation between the absorption - depth ratio of @xmath35972 to @xmath36355 ) and the brightness decline rate .
wang et al .
( 2009 ; 2013 ) and foley et al .
( 2011a ) separated sne ia into high - velocity and normal - velocity groups with a boundary at 11,800kms@xmath5 at peak brightness , and found that the former are @xmath11mag ( on average ) redder in @xmath12 than the latter .
meanwhile , high - resolution spectral observations provide a powerful way to study absorption along the line of sight , both from the interstellar medium and circumstellar material .
patat et al . (
2007 ) found a complex of d lines that showed evolution in sn 2006x ( wang et al .
two additional cases of time - variable na absorption are provided by blondin et al .
( 2009 ) and simon et al .
sternberg et al . ( 2014 ) found that in their sample with high - resolution spectra , @xmath13% of sne ia exhibit time - variable na , indicating the presence of circumstellar material and suggesting that it may be more common than expected in sne ia , though some objects do not show evolution ( e.g. , sn 2014j ; graham et al .
2015 ) .
spectropolarimetry can be used to probe the geometry of sne ia ( see wang & wheeler 2008 for a review ) .
the continuum polarization , an indication of the photosphere s shape , was found to be quite low in sne ia , on the order of a few tenths of a percent ( hflich 1991 ; wang et al .
but for individual sne ia , significant line polarization is sometimes observed ( e.g. , wang et al .
2003 ; kasen et al .
wang et al . ( 2007 ) also found a correlation between the degree of polarization of @xmath36355 and the brightness decline rate .
observationally , there are numerous efforts to discover sne ia at very early times , which can benefit follow - up observations in many ways
. recent examples of early - observed and well - studied sne ia include sn 2009ig ( foley et al .
2012 ) , sn 2011fe ( nugent et al . 2011 ; li et al . 2011 ) , sn 2012cg ( silverman et al .
2012a ) , sn 2013dy ( zheng et al . , 2013 ) , iptf13ebh ( hsiao et al . 2015 ) , sn 2014j ( zheng et al .
2014 ; goobar et al .
2014 ; graham et al .
2015 ) , and asassn-14lp ( shappee et al . 2016 ) ; like sn 2016coj discussed here , they were either discovered or detected shortly after exploding . in 2011 , the observing strategy for our lick observatory supernova search ( loss ; filippenko et al . 2001
; filippenko 2005 ; leaman et al .
2011 ) with the 0.76 m katzman automatic imaging telescope ( kait ) was modified to monitor fewer galaxies but at a more rapid cadence , with the objective of promptly identifying very young sne ( hours to days after explosion ) . in the past few years , this strategy has led to discoveries of all types of young sne , where we define a sn to be `` young '' if there was a kait nondetection 13 days before the first detection or if it was spectroscopically confirmed to be within a few days after explosion . sn 2012cg ( silverman et al .
2012a ) was the first case , followed by more than a dozen others ( e.g. , sn 2012ck , kandrashoff et al .
2012 ; sn 2012ea , cenko et al .
2012 ; sn 2013ab , blanchard et al .
2013 ; sn 2013dy , zheng et al .
2013 ; sn 2013ej , dhungana et al .
2016 ; sn 2013gh , hayakawa et al .
2013 ; sn 2013fv , kim et al . 2013a ; sn 2013gd , casper et al .
2013 ; sn 2013gy , kim et al . 2013b ; sn 2014c , kim et al . 2014a ; sn 2014j , though not discovered by kait , but with kait early detections , see zheng et al .
2014 ; sn 2014ce , kim et al . 2014b ; sn 2014eh , kumar et al .
2014 ; sn 2015n , stegman et al .
2015a ; sn 2015u , shivvers et al .
2016 ; sn 2015x , hughes et al . 2015 ; sn 2015o , ross et al .
2015 ; sn 2015be , stegman et al . 2015b
; and sn 2016esw , halevi et al . 2016 ; ) .
sn 2016coj was another sn discovered by kait when very young , merely @xmath14day after the fitted first - light time ( fflt ) . here
we present the first 40 days of our optical photometric , low- and high - resolution spectroscopic , and spectropolarimetric follow - up observations and analysis of it .
sn 2016coj was discovered in an 18 s unfiltered kait image taken at 04:39:05 on 2016 may 28 ( ut dates are used throughout this paper ) , at @xmath15mag ( close to the @xmath16 band ; see li et al .
it was reported to the transient name server ( tns ) shortly after discovery by yuk , zheng , & filippenko ( see also zheng et al .
we measure its j2000.0 coordinates to be @xmath17 , @xmath18 , with an uncertainty of @xmath19 in each coordinate .
sn 2016coj is @xmath20 east and @xmath21 north of the nucleus of the host galaxy ngc 4125 , which has redshift @xmath0 according to the nasa / ipac extragalactic database ( ned ) , an early - type peculiar elliptical morphology ( e6 pec ; de vaucouleurs et al .
1991 ) , and a stellar mass of @xmath22m@xmath23 from its 3.6@xmath24 flux ( wilson et al .
2013 ) .
kait performed photometric follow - up observations of sn 2016coj with nearly daily cadence after discovery .
the data were reduced using our image - reduction pipeline ( ganeshalingam et al .
we applied an image - subtraction procedure to remove host - galaxy light , and point - spread - function photometry was then obtained using daophot ( stetson 1987 ) from the idl astronomy user s library .
the unfiltered instrumental magnitudes , which are found to be close to the @xmath16 band ( li et al .
2003 ) , are calibrated to local sdss standards ( see figure [ fig_kaitclear_findingchart ] ) transformed into landolt @xmath16-band magnitudes . here
we publish our unfiltered photometry ( table 1 ) .
we have also obtained a filtered data sequence , but we are still awaiting high - quality galaxy template images in those bands .
interestingly , we find that sn 2016coj was detected in a kait prediscovery image taken at 04:30:35 on may 24 ( see middle panels of figure [ fig_findingchart_subtracted ] ) with an unfiltered mag of @xmath25 , which means the sn had brightened @xmath26 mag in the following four days until it was discovered .
in addition , an unfiltered prediscovery detection was obtained at 21:48:57 on may 23 , 6.7 hr earlier than kait s first detection , by r. arbour with a 0.35 m @xmath27 schmidt - cassegrain reflector ( see left panels of figure [ fig_findingchart_subtracted ] ) . using a template image taken on 2016 april 5 , we performed the same subtraction and calibration methods as for the kait unfiltered images .
we find a sn unfiltered brightness of @xmath28mag , consistent with kait s first detection . a @xmath29mag detection before peak magnitude , along with our analysis in the following section , confirms that sn 2016coj is one of the youngest sne ia ever detected .
lcccc 0.4 mm + 57520.2370 & may 12.2370 & @xmath3019.6 & - & kait + 57522.2513 & may 14.2513 & @xmath3019.4 & - & kait + 57524.2240 & may 16.2240 & @xmath3019.4 & - & kait + 57525.2479 & may 17.2479 & @xmath3019.3 & - & kait + 57527.2708 & may 18.2708 & @xmath3019.2 & - & kait + 57531.9092 & may 23.9092 & 18.06 & 0.42 & r. arbour + 57532.1877 & may 24.1877 & 18.02 & 0.22 & kait + 57536.2694 & may 28.2694 & 14.98 & 0.03 & kait + 57537.2196 & may 29.2196 & 14.50 & 0.04 & kait + 57538.1827 & may 30.1827 & 14.11 & 0.03 & kait + 57539.1814 & may 31.1814 & 13.85 & 0.03 & kait + 57540.2518 & june 01.2518 & 13.54 & 0.04 & kait + 57541.2022 & june 02.2022 & 13.39 & 0.06 & kait + 57542.2075 & june 03.2075 & 13.18 & 0.03 & kait + 57543.2120 & june 04.2120 & 13.16 & 0.03 & kait + 57544.1995 & june 05.1995 & 13.02 & 0.03 & kait + 57545.2245 & june 06.2245 & 13.00 & 0.03 & kait + 57546.2200 & june 07.2200 & 12.95 & 0.03 & kait + 57547.2051 & june 08.2051 & 12.93 & 0.03 & kait + 57548.2197 & june 09.2197 & 12.94 & 0.04 & kait + 57549.2162 & june 10.2162 & 12.99 & 0.05 & kait + 57550.2638 & june 11.2638 & 12.97 & 0.02 & kait + 57551.2201 & june 12.2201 & 13.01 & 0.03 & kait + 57552.2516 & june 13.2516 & 13.03 & 0.03 & kait + 57553.2115 &
june 14.2115 & 13.15 & 0.03 & kait + 57555.2189 & june 16.2189 & 13.27 & 0.03 & kait + 57556.2284 & june 17.2284 & 13.30 & 0.03 & kait + 57558.2110 & june 19.2110 & 13.52 & 0.02 & kait + 57559.2229 & june 20.2229 & 13.61 & 0.04 & kait + 57560.2189 & june 21.2189 & 13.66 & 0.04 & kait + 57561.2097 & june 22.2097 & 13.70 & 0.04 & kait + two classification spectra of sn 2016coj were obtained shortly ( @xmath31hr ) after the sn was discovered ( @xmath32 5.0 days after the fflt ) .
the spectra were taken with the kast double spectrograph ( miller & stone 1993 ) on the shane 3 m telescope at lick observatory and the floyds robotic spectrograph on the las cumbres observatory global telescope network ( lcogt ; brown et al .
2013 ) 2.0 m faulkes telescope north on haleakala , hawaii .
we obtained nearly daily spectra of sn 2016coj with different instruments including kast , floyds , the bfosc spectrograph on the 2.16 m telescope at xinglong station of naoc ( china ) , the low resolution imaging spectrometer ( lris ; oke et al .
1995 ) on the 10 m keck i telescope , and the kitt peak ohio state multi - object spectrograph ( kosmos ; martini et al .
2014 ) on the kpno mayall 4 m telescope .
data were reduced following standard techniques for ccd processing and spectrum extraction using iraf .
the spectra were flux calibrated through observations of appropriate spectrophotometric standard stars .
all kast and lris spectra were taken at or near the parallactic angle ( filippenko 1982 ) to minimize differential light losses caused by atmospheric dispersion .
low - order polynomial fits to calibration - lamp spectra were used to calibrate the wavelength scale , and small adjustments derived from night - sky lines in the target frames were applied .
flux calibration and telluric - band removal were done with our own idl routines ; details are described by silverman et al .
( 2012d ) .
we also obtained four epochs of lick / shane spectropolarimetry using the polarimetry mode of the kast spectrograph on may 30 , june 8 , june 16 , and july 6 .
the spectra were observed at each of four waveplate angles ( @xmath33 , @xmath34 , @xmath35 , and @xmath36 ) with several waveplate sequences coadded to improve the signal - to - noise ratio ( s / n ) .
each night , both low- and high - polarization standard stars were also observed in order to calibrate the data .
all of the spectropolarimetric reductions and calculations follow the method described by mauerhan et al .
( 2015 ) , and the polarimetric parameters are defined in the same manner ( stokes parameters @xmath37 and @xmath38 , debiased polarization @xmath39 , and sky position angle @xmath40 ) .
in addition , we observed sn2016coj on may 31 , june 2 , 4 , and 6 with the 2.4 m automated planet finder ( apf ) telescope at lick observatory . the apf hosts the levy spectrograph , a high - resolution optical echelle spectrograph with resolution @xmath16(5500 ) @xmath41 110,000 with a slit width of 1 ( vogt et al . 2014 ) . at each epoch we obtained three 1800s spectra with
the m decker ( 10 wide , 80 long to allow for background subtraction ) , reduced the data with a custom pipeline , and corrected for the redshift of the host galaxy ( @xmath0 ) and for the barycentric velocity ( @xmath42kms@xmath5 ) . because the apparent magnitude ( peaking at @xmath43 mag ) of sn2016coj is a bit faint for apf ,
the s / n of our spectra was @xmath44 at best , significantly lower than obtained with apf spectra of the bright ( peak @xmath45mag ) , nearby snia 2014j ( graham et al .
2015 ) .
model fit to the red solid circles .
the blue cross marks the earliest detection at @xmath2d after the fitted first light observed by r. arbour .
red triangles show the kait upper limits before explosion . ]
figure [ fig_lcfit ] shows our unfiltered light curve of sn 2016coj . in order to determine the first - light time @xmath46 ( note that the sn may exhibit a dark phase " )
, one can assume that the sn luminosity scales as the surface area of the expanding fireball , and therefore increases quadratically with time ( @xmath47 , commonly known as the @xmath48 model ; arnett 1982 ; riess et al .
the @xmath48 model fits well for several sne ia with early - time observations ( e.g. , sn 2011fe , nugent et al .
2011 ; sn 2012ht , yamanaka et al .
some studies also adopt a @xmath49 model ( @xmath50 varies from @xmath51 to @xmath26 ; e.g. , conley et al .
2006 ; ganeshalingam et al .
2011 ; firth et al .
interestingly , zheng et al .
( 2013 , 2014 ) use a broken - power - law model to estimate the first - light time of sn 2013dy and sn 2014j .
however , since our early - time photometric coverage of sn 2016coj is not as good as that of sn 2013dy and sn 2014j , we simply apply the @xmath48 model to fit the kait unfiltered data for the first few days ( red solid circles in figure [ fig_lcfit ] ) ; thereafter , the light curve starts deviating from the @xmath48 model .
we also exclude arbour s unfiltered detection ( blue cross ) , considering the different response curve compared to the kait unfiltered data : arbour s unfiltered band is closer to @xmath52 ( see botticella et al .
2009 ) , while kait s is closer to @xmath16 ( see li et al .
2003 ) .
we find that the best @xmath48 model fit gives the first - light time to be mjd @xmath53 , around may 23.33 . here the uncertainty ( not including the dark phase " ) is estimated by calculating the reduced @xmath54 ratio with the minimum reduced @xmath54 at 90% confidence level , when @xmath46 changes around the best - fitted value while all the other parameters are fixed with the best - fitted value .
note that the uncertainty does not include any systematic error caused by the @xmath48 model fitting .
for example , if we include ( or exclude ) one data point before and after the dataset we used , the best - fit first - light time deviates @xmath55 to 1.0 days from the above first - light time .
therefore , there could be a systematic error of up to 1.0d from this method , which we did not include in the following analysis .
our results show that the first detection ( from an image by r. arbour ) was merely @xmath2d after first light , or 0.9d from kait s first detection on may 24 .
this makes sn 2016coj one of the earliest detected sne ia slightly later than sn 2013dy ( @xmath56hr after first light ; zheng et al .
2013 ) and sn 2011fe ( @xmath57hr after first light ; nugent et al .
2011 ) , but similar to sn 2009ig ( @xmath58hr after first light ; foley et al . 2012 ) . applying a low - order polynomial fit
, we find that sn 2016coj reached a peak magnitude of @xmath59 at mjd @xmath60 in kait unfiltered data .
although we do not present @xmath1-band data because no @xmath1-band template image is currently available , the fit allows us to determine the @xmath1-band peak time : mjd @xmath61 , similar to the result with unfiltered data .
this means sn 2016coj was discovered only 4.9d after the fitted first light , or 11.1d before maximum light .
the distance modulus of the host galaxy ngc 4125 is quite uncertain owing to different measurements given in ned .
however , some of them are outdated , or adopted an inappropriate h@xmath62 value .
the one with the smallest uncertainty ( and also the latest estimate ) is @xmath63mag ( tully et al .
2013 ) , which was based on h@xmath64kms@xmath5mpc@xmath5 , quite close to the current widely accepted value of h@xmath65 .
we therefore adopt this distance for the following ananlysis . with @xmath66mag ( schlafly et al . 2011 ) and
very small ( even negligible ) host - galaxy extinction ( see [ ss : spectra ] and [ ss : highresspectra ] ) , this implies sn 2016coj has @xmath67mag at peak brightness .
our preliminary measurement of @xmath1-band data , assuming the host background contamination is small , shows that the @xmath1-band peak is @xmath68mag and @xmath69mag .
this gives @xmath70mag , but we expect @xmath71mag from the phillips ( 1993 ) relation with the above value of @xmath72 ; thus , sn 2016coj is a normal - brightness sn ia .
its @xmath69mag is also typical of normal sne ia ( see also [ ss : classification ] for the spectral classification ) .
we obtained optical spectra of sn 2016coj nearly daily for a month ( fig .
[ fig_spectra ] ) , sometimes obtaining multiple spectra in a given night .
we first examine the d absorption feature , which is often converted into reddening but with large scatter over the empirical relationship ( poznanski et al .
2011 ) , in several of our high - s / n spectra .
the absorption is not clearly detected at both the d rest - frame wavelength and the redshifted wavelength of sn 2016coj .
however , there appears to be a weak absorption feature consistent with the rest - frame d wavelength .
if real , this could be caused by a milky way component , which has @xmath66mag according to schlafly et al .
since we do not detect similar absorption at the redshifted wavelength of sn 2016coj , we can put an upper limit of @xmath73mag of host - galaxy extinction .
however , if the weak absorption feature is caused by noise instead of milky way gas , we can determine an upper limit on @xmath74 through comparison with our spectra of sn 2013dy ( zheng et al .
2013 ) , where we clearly detect the d absorption with the same instrument setting . for sn 2013dy , the equivalent width ( @xmath75 )
is @xmath76from both the milky way and host galaxy , giving @xmath77mag .
our similar - quality data on sn 2016coj should allow a detection of 1/3 ( or less ) of d absorption if it exists , yielding an upper limit of @xmath78mag of host - galaxy extinction .
lastly , we also estimate a @xmath79 upper limit on the @xmath75 of an undetected feature in a spectrum using the equation presented by leonard ( 2007 ) : @xmath80 where @xmath81 is the spectral resolution element ( in ) , @xmath82 is the 1@xmath83 root - mean - square fluctuation of the flux around a normalized continuum level , @xmath84 is the full - width at half - maximum intensity ( fwhm ) of the expected line feature , and @xmath1 is the number of bins per resolution element . for our high - s / n kast spectra , we measure @xmath85 , @xmath86 , @xmath87 , and @xmath88 , which gives @xmath89 , and @xmath90mag of host - galaxy extinction .
all of the above suggests that the host - galaxy extinction of sn 2016coj is likely to be very small , consistent with the nondetection of d absorption in our high - resolution spectra ( see [ ss : highresspectra ] ) .
however , note that since the d vs. extinction relation has large scatter , even a nondetection of d does not fully exclude the possibility that there may be some dust along the sn line of sight . the spectra show absorption features from ions typically seen in sne ia including , , , , , and .
we do not find a clear feature ( e.g. , zheng et al .
2013 ) , which is found in over one - fourth of all sne ia ( e.g. , parrent et al . 2011
; thomas et al . 2011 ; folatelli et al .
2012 ; silverman et al .
strong absorption features of , including @xmath34000 , @xmath35972 , and @xmath36355 , are clearly seen in all spectra . the @xmath35972 feature in sn
2016coj is quite strong relative to those in sn 2012cg and sn 2013dy , though it is relatively small if compared with a large sn ia sample ( see silverman et al .
2012c ) .
we measure the individual line velocities from the minimum of the absorption features ( see silverman et al .
2012c , for details ) and show them in figure [ fig_spec_velocity ] .
the velocities all features decrease from @xmath9115,000kms@xmath5 at discovery to @xmath9213,000kms@xmath5 around maximum light , and they continue to decrease thereafter . -band maximum light . ]
in addition to the usual photospheric - velocity feature ( pvf ) of h&k , sn 2016coj exhibits a high - velocity feature ( hvf ; e.g. , mazzali et al .
2005 ; maguire et al . 2012 ; folatelli et al . 2013 ; childress et al .
2014 ; maguire et al . 2014 ; silverman et al .
2015 ) in nearly all of the early - time spectra .
this hvf appears to be detached from the rest of the photosphere , with a velocity of @xmath93kms@xmath5 at discovery and slowing down to @xmath94kms@xmath5 at @xmath95d after the fitted first - light time .
the hvf feature of h&k stays for a long time , being distinct until roughly age + 11d ; thereafter , it is a high - velocity shoulder of the h&k absorption .
a near - infrared ( nir ) triplet hvf is also found in the first few spectra that covered the wavelength range before maximum light , and the velocity of @xmath96kms@xmath5 is in good agreement with that of the h&k hvf at early times , though it is slightly smaller in the first - epoch spectrum .
such hvfs are seen in a few other well - observed sne ia , including sn 2005cf ( wang et al . 2009 ) and sn 2012fr ( e.g. , maund et al .
2013 ; childress et al . 2013 ; zhang et al .
however , in sn 2016coj , the nir triplet hvf becomes weaker around peak brightness , and it completely disappears @xmath95d later and thereafter .
this is different from the h&k hvf , which is seen for a much longer time .
it is not obvious why the hvf of the nir triplet goes away after peak brightness while the hvf of h&k persists .
one possibility is that the apparent hvf of h&k after peak could actually be @xmath33858 ( e.g. , foley 2013 ) .
in fact , it is possible that the early - time apparent hvf of h&k could be a mixture of @xmath33858 ( including both the hvf and pvf ) plus the true hvf of h&k .
if so , the velocity of the h&k hvf could be smaller than that shown in figure [ fig_spec_velocity ] , and thus more consistent with the velocity of the nir triplet hvf , but this case is too complicated to verify .
one note about the triplet feature is that we adopted only one component in our fit .
however , our early - time spectra before peak brightness reveal that the triplet has a double absorption profile . following the zhao et al .
( 2016 ) method to fit the triplet with both hvf and pvh ( zhao et al . also adopted a second , faster hvf , but that is not clear in sn 2016coj ) , we find an hvf triplet velocity of @xmath97kms@xmath5 and a pvf triplet velocity of @xmath98kms@xmath5 . the hvf velocity is smaller than that of both h&k and the nir triplet .
if the hvf really exists in the triplet , it suggests that the oxygen in the outer layers is not completely burned ( see zhao et al .
the strong absorption of si ii @xmath36355 is commonly used to estimate the photospheric velocity .
as shown in figure [ fig_spec_velocity ] , the si ii @xmath36355 velocity of sn 2016coj decreases rapidly from @xmath99kms@xmath5 at discovery to @xmath100kms@xmath5 around peak brightness , and then slowly decreases to @xmath101kms@xmath5 at + 11.0d after peak . a velocity of @xmath100kms@xmath5 at peak brightness is @xmath102kms@xmath5higher than average in sne ia ( e.g. , wang et al . 2013
; @xmath103 away from the mean of their sn ia velocity distribution fitted with a gaussian ) .
here , we compare the photospheric velocity measurement of sn 2016coj with the three well - observed sne ia 2009ig ( foley et al . 2012 ; marion et al . 2013 ) , 2012cg ( silverman et al .
2012a ) , and 2013dy ( zheng et al .
note that while both sn 2012cg and sn 2009ig have an hvf identified for @xmath36355 , we consider only the photospheric component .
figure [ fig_spec_velocity ] displays the photospheric velocity evolution over time for the four sne ia .
overall , the photospheric velocity evolution is similar to the evolution seen in most sne ia ( e.g. , benetti et al . 2005 ; foley et al . 2011b ; silverman et al .
2012c ) : the velocity drops rapidly at early times ( within the first week after explosion ) , and then slowly but steadily decreases thereafter . for each of these four sne
, we consequently try to fit the early - time velocities ( typically within 10d after first light ) with a power - law function , @xmath104 , where @xmath105 is the time after first light ( @xmath46 ) ; the results are shown in the top panel of fig .
[ fig_spec_velocity ] for each sn .
this is very similar to the method silverman et al .
( 2015 , fig .
12 ) adopted , but they used a natural exponential function to fit the velocities before + 5d after peak brightness and also obtained reasonable fitting results .
in fact , piro & nakar ( 2013 , eq . 13 ) mathematically show that the photospheric velocity could decay as a power law at early times . for the later velocities ( typically @xmath106d after first light )
, we then fit them with a linear function , @xmath107 ( results shown in the middle - top panel ) ; silverman et al .
( 2012c , fig . 5 ) also use the same method to fit their data around peak brightness . as seen in figure [ fig_spec_velocity ]
, both the power - law function and the linear function can fit the corresponding data well , but only in their respective regimes early - time data for the power - law function and later - time data for the linear function . as with the early - time light - curve ( zheng et al .
2013 , 2014 ) , we find that a broken - power - law function is useful for fitting the photospheric velocity evolution ; a low - index power - law function approximates the linear function found at late times .
specifically , @xmath108}^{-1/s } , \label{formula_bkn_vc}\ ] ] where @xmath109 is the photospheric velocity , @xmath110 is a scaling constant , @xmath105 is the time after first light ( @xmath46 ) , @xmath111 is the break time , @xmath112 and @xmath113 are the two power - law indices before and after the break ( respectively ) , and @xmath114 is a smoothing parameter .
we apply this broken - power - law function to the entire dataset of photospheric velocities for all four sne until about a month after the explosion . our fitting results ( we fixed @xmath114 to be @xmath115 ) are listed in table 2 and shown in the middle - bottom panels in figure [ fig_spec_velocity ] .
lc|c|ccc 0.4 mm sn & power law & linear@xmath116 & + sn 2009ig & [email protected] & -32@xmath11731 & [email protected] & [email protected] & 0.04 + sn 2012cg & [email protected] & -77@xmath11732 & [email protected] & [email protected] & 0.04 + sn 2013dy & [email protected] & -54@xmath11732 & [email protected] & [email protected] & 0.14 + sn 2016coj & [email protected] & -95@xmath1173 & [email protected] & [email protected] & 1.72 + the power - law indices from both the power - law fitting ( @xmath118 ) and broken - powerlaw fitting ( @xmath119 ) at early times are consistent with the value of @xmath120 adopted by piro & nakar ( 2014 ) when fitting three sne ia ( sne 2009ig , 2011fe , and 2012cg ) , and are also the value adopted by shappee et al .
( 2016 ) when fitting asassn-14lp .
the index from the broken power law ( @xmath119 ) is slightly steeper than that from the power law ( @xmath118 ) . at late times
( around maximum light ) with linear fitting , the rate of velocity decrease from the fitting is slightly larger than the average rate of @xmath121kms@xmath5d@xmath5 found by silverman et al .
( 2012c ) for a large sample of sne ia .
overall , the broken - power - law function can fit the photospheric velocity evolution well for all four sne until a month after explosion ( see the small residuals at the bottom panel of figure [ fig_spec_velocity ] and the reduced @xmath54 given in table 2 ) .
this function also has the potential to fit the photospheric velocity evolution of most other sne ia as well , given that most sne ia have very similar velocity evolution ( e.g. , silverman et al .
2012b , 2012c ) .
high - cadence spectroscopy is required to verify this , especially at early times .
however , currently it remains unclear whether there is a good physical explanation behind the fitting ; piro & nakar ( 2013 ) show that the photospheric velocity could decay as a power law at early times , but our broken - power - law function fitting extends to a much later time .
we use the supernova identification code ( snid ; blondin & tonry 2007 ) to spectroscopically classify sn 2016coj . for nearly all of the spectra presented here ,
we find that sn 2016coj is spectroscopically similar to many normal sne ia . compared to sn 1992a ( @xmath122mag and @xmath123mag ; della valle et al .
1998 ) and sn 2002er ( @xmath124mag and @xmath125mag ; pignata et al .
2004 ) , for example , sn 2016coj has similar spectra , absolute magnitude , and @xmath72 .
another spectroscopic comparison is the so - called ratio , @xmath126 ( ) ( the ratio of @xmath35972 to @xmath36355 ) , defined by nugent et al . ( 1995 ) using the _ depths _ of spectral features and later by hachinger et al .
( 2006 ) using their pseudo - equivalent widths .
hachinger et al .
( 2006 ) found a good correlation between the @xmath126 ( ) and @xmath72 ( see their figure 13 ) .
we measure sn 2016coj to have @xmath126 ( ) @xmath127 , with @xmath128mag , placing sn 2016coj in the normal sn ia region in figure 13 of hachinger et al .
( 2006 ) , very close to sn 2002er .
thus , we conclude that sn 2016coj is a spectroscopically normal sn ia , consistent with the photometric analysis given in [ ss : lightcurves ] . ) , with several major features labeled .
bottom panel : position angle ( @xmath40 ) for the corresponding epochs .
@xmath40 is underdetermined where @xmath129 ; those points are omitted . ] the interstellar polarization ( isp ) appears to be low in the direction of sn 2016coj .
indeed , the estimated value of @xmath130mag indicates that the extinction from the milky way and host galaxy are not substantial ; a small contribution from isp is thus to be expected . according to serkowski et al .
( 1975 ) , an upper limit to the isp is given by @xmath131 , which implies @xmath132 for sn 2016coj . to obtain a direct estimate of the galactic component of isp , we observed three galactic stars in the vicinity of the sn position : hd104436 ( a3v ) , hd106998 ( a5v ) , and hd108907 ( m3iii ) .
we measure respective @xmath52-band polarization and @xmath40 values of @xmath133 and @xmath134 . under the reasonable assumption of low intrinsic polarization for these stars , the resulting average values of @xmath135% ,
@xmath136 confirm the low galactic polarization .
furthermore , the lack of d absorption lines in our low- and high - resolution spectra ( see [ ss : spectra ] and [ ss : highresspectra ] ) indicates low extinction from the early - type host galaxy , and thus implies that the host isp is probably even lower than the small galactic value .
the instrumental polarization of the kast instrument is also low .
measurements of the low - polarization standard star bd@xmath137332642 at each epoch indicate an average @xmath52-band polarization of @xmath138% , with a standard deviation of 0.05% between all four epochs ; the average value is consistent with that reported by schmidt et al .
( 1992 ) for this star , which indicates that the low level is intrinsic to the source and that kast contributes an insignificant amount of instrumental polarization to the measurements .
the standard deviation is near the systematic uncertainty level we typically experience using the spectropolarimetry mode of kast .
our observations therefore constrain the average instrumental polarization to @xmath139 .
based on the low values of isp and instrumental polarization , we move forward without attempting to subtract their minor contributions from the data . our spectropolarimetry results are shown in figure [ fig_specpol ] and the integrated broadband measurements are listed in table3 . on day 6.9 ,
the source exhibits weak polarization in the continuum at a level of @xmath140% , integrated over the wavelength range 67007150 .
this is consistent with the weak levels of continuum polarization that are typically associated with sne ia ( wang & wheeler 2008 ) , although we note that some fraction of the polarization , perhaps half , could potentially be contributed by isp .
strong polarization is exhibited across prominent line features , particularly @xmath36355 and the nir triplet , at levels of @xmath141% and @xmath142% , respectively .
the polarization feature appears to exhibit two peaks , perhaps associated with the high- and low - velocity components .
the position angles across the polarized line features , particularly , are close to that of the continuum , which suggests an axisymmetric configuration for the sn . by day 16.0
, the continuum polarization is consistent with having no change relative to day 6.9 , while has increased in strength slightly to peak at this epoch .
a gaussian fit to the feature indicates a line polarization of @xmath143 with respect to the continuum level . for polarization ,
the enhancement of the high - velocity component from day 6.9 has disappeared and the peak of the lower - velocity component has increased by @xmath140% . by day 24.0 , the continuum and line polarization appears to have dropped substantially for wavelengths shortward of 7000 , with no significant change apparent at longer wavelengths ; polarization in the continuum region is undetected at this epoch .
if real , such a continuum polarization drop roughly one week after peak would be reminiscent of the evolution of sn2001el ( wang et al .
however , by day 44.0 the continuum polarization appears to have regained the strength exhibited on day 16.0 and earlier .
has restrengthened as well , while declining in radial velocity along with the minimum of the weakening absorption profile .
based on this unexpected restrengthening , we exercise caution regarding the temporarily weakened polarization on day 24.0 , as we are concerned that this could be the result of a systematic error .
the drop in polarization appears to have only affected the stokes @xmath37 parameter ( derived from exposures with polarimeter waveplate angles at 0@xmath144 and 45@xmath144 ) .
each of our three @xmath37 sequences of the sn are consistent , and we see no such change in the @xmath37 parameter of our standard - star observations from the same night . thus , if the change on day 24.0 is the result of systematic error ( e.g. , some unknown temporary source of instrumental polarization above our typical limit of @xmath145% ) , then it must have occurred over an hourly timescale . alternatively , a subsequent rise in continuum polarization on day 44.0 could result from the appearance of weak line features in the our chosen continuum region ( 67007150 ) , but in this case we would not expect the simultaneous rise in the feature . as a final possibility , the temporary influence of a separate light - echo component , possibly associated with dust in the host ism , could result in the observed fluctuation ; this possibility has the advantage of accounting for the brief change in the continuum and line polarization simultaneously , and it would also explain why the reddest wavelengths are not significantly affected .
overall , the spectropolarimetric character of sn2016coj is consistent with the trends exhibited by normal " sne ia .
for example , maund et al . (
2010 ) reported a correlation between the polarization of the @xmath36355 feature , measured near or before peak luminosity , and the radial - velocity decline rate of the absorption minimum ( also see leonard et al . 2006 ) , physically interpreted as evidence for a single geometric configuration for normal sne ia . at peak brightness on day 16.0
, the line polarization of @xmath143 combined with our measured value of @xmath146kms@xmath5day@xmath5 for the velocity evolution , shows that sn 2016coj falls where expected on the correlated distribution of sne ia reported by maund et al .
( 2010 ) , and within the range of high - velocity explosions .
.polarization of sn 2016coj [ cols="<,^,^,^,^",options="header " , ] [ hstspitzer_model ] we examine the apf high - resolution spectra for narrow absorption features , such as those that were identified in apf spectra of sn 2014j ( graham et al .
we began spectral monitoring with the apf based on an early classification and the assumption of a host - galaxy distance smaller than that adopted here .
the object s peak apparent brightness ended up being @xmath147mag fainter than that of sn2014j , and fainter than the projected minimum we typically require for triggering the apf .
for this reason , the s / n of our sn2016coj apf spectra is quite low .
instead of ceasing our apf monitoring we obtained multiple observations over several nights in order to stack our spectra , but ultimately we do not identify any absorption features of d @xmath1485889.95 , 5895.92 , h&k @xmath1483933.7 , 3968.5 , @xmath1487664.90 , 7698.96 , h@xmath118 @xmath36562.801 , h@xmath149 @xmath34861.363 , or the diffuse interstellar bands ( @xmath150 , 5797 , 6196 , 6283 , 6613 ) . since the d feature is most useful for constraining the presence of circumstellar material and line - of - sight host - galaxy dust extinction , and owing to grating blaze
is in a region of relatively higher s / n ( @xmath45 ) , we estimate an upper limit on its @xmath75 in the following way . the flux of the continuum - normalized stacked apf spectra in the region of d , shown in black in figure [ fig_spec_high_resolution ] , has a standard deviation of @xmath151 . as an upper limit on the depth of an absorption feature that we could have detected
, we use @xmath152 . our instrumental configuration for the levy spectrograph results in a spectral resolution of @xmath153 , from which we estimate that the minimum fwhm of a detected feature is @xmath154 .
assuming a gaussian profile for this hypothetical absorption , we constrain the d feature to have @xmath155 . based on figure 9 of phillips et al .
( 2013 ) , this puts an upper limit on host - galaxy extinction of @xmath156mag ( with @xmath157mag assuming @xmath158 ) .
although this is a rather large upper limit , it is consistent with the small host - galaxy extinction constrained from our low - resolution spectra ( see [ ss : spectra ] ) and also with the low extinction expected given the early type of the host , ngc 4125 .
in this paper we have presented optical photometric , low- and high - resolution spectroscopic , and spectropolarimetric observations of sn 2016coj , one of the youngest discovered and best - observed sne ia .
our clear - band light curve shows that our first detection is merely @xmath2d after the fitted time of first light , making it one of the earliest detected sne ia .
we estimate that sn 2016coj took @xmath159 d after the fitted first - light time to reach @xmath1-band maximum .
its maximum brightness has a normal luminosity , @xmath160mag .
an estimated @xmath72 value of 1.25mag along with spectral information support its normal sn ia classification . in the well - observed low - resolution spectral sequence ,
we identify a high - velocity feature from both h&k and the nir triplet , and also possibly from the triplet .
sn 2016coj has a si ii @xmath36355 velocity of @xmath100kms@xmath5 at peak brightness , @xmath161kms@xmath5 higher than that of typical sne ia .
we find that the si ii @xmath36355 velocity decreases rapidly during the first few days and then slowly decreases after peak brightness , very similar to that of other sne ia .
a broken - power - law function can well fit the si ii @xmath36355 velocity for up to about a month after first light .
we estimate there to be very small host - galaxy extinction based on the lack of na i d lines from the host galaxy in our low- and high - resolution spectra .
our four epochs of spectropolarimetry show that sn 2016coj exhibits weak polarization in the continuum , but the line polarization is quite strong ( @xmath162% ) at peak brightness .
a.v.f.s group at u.c .
berkeley is grateful for financial assistance from national science foundation ( nsf ) grant ast-1211916 , gary & cynthia bengier , the richard & rhoda goldman fund , the christopher r. redlich fund , and the tabasgo foundation .
is supported by an nsf astronomy and astrophysics postdoctoral fellowship under award ast-1302771 .
uc irvine observing runs were supported in part by nsf grant ast-1412693 .
some of the data presented herein were obtained using the uc irvine remote observing facility , made possible by a generous gift from john and ruth ann evans .
v.n.b . and m.c .
gratefully acknowledge assistance from an nsf research at undergraduate institutions ( rui ) grant ast-1312296 .
the ucsc group is supported in part by nsf grant ast-1518052 , and by fellowships from the alfred p. sloan foundation and the david and lucile packard foundation to r.j.f .
observations by the ucla group are supported by nsf grant ast-1412315 to t.t .
is financially supported by the national science foundation of china ( nsfc grants 11178003 and 11325313 ) .
this work was partially supported by the open project program of the key laboratory of optical astronomy , national astronomical observatories , chinese academy of sciences .
, c.m . , and d.a.h .
are supported by nsf grant ast-1313484 .
we thank the staffs of the various observatories at which data were obtained .
some of the data presented herein were obtained at the w. m. keck observatory , which was made possible by the generous financial support of the w. m. keck foundation .
based in part on observations obtained at kitt peak national observatory , national optical astronomy observatory ( noao prop .
i d 2015b-0313 ; pi foley ) , which is operated by the association of universities for research in astronomy ( aura ) , inc .
, under cooperative agreement with the nsf .
the authors are honored to be permitted to conduct astronomical research on iolkam duag ( kitt peak ) , a mountain with particular significance to the tohono oodham .
this work makes use of observations from the lcogt network .
we thank lauren weiss and bradford holden for obtaining apf data at lick observatory .
research at lick observatory is partially supported by a generous gift from google .
|
the type ia supernova ( sn ia ) 2016coj in ngc 4125 ( redshift @xmath0 ) was discovered by the lick observatory supernova search 4.9 days after the fitted first - light time ( fflt ; 11.1 days before @xmath1-band maximum ) .
our first detection ( pre - discovery ) is merely @xmath2 day after the fflt , making sn 2016coj one of the earliest known detections of a sn ia .
a spectrum was taken only 3.7 hr after discovery ( 5.0 days after the fflt ) and classified as a normal sn ia .
we performed high - quality photometry , low- and high - resolution spectroscopy , and spectropolarimetry , finding that sn 2016coj is a spectroscopically normal sn ia , but with a high velocity of @xmath36355 ( @xmath4kms@xmath5 around peak brightness )
. the @xmath36355 velocity evolution can be well fit by a broken - power - law function for up to a month after the fflt .
sn 2016coj has a normal peak luminosity ( @xmath6 mag ) , and it reaches a @xmath1-band maximum @xmath7 16.0 d after the fflt .
we estimate there to be low host - galaxy extinction based on the absence of na i d absorption lines in our low- and high - resolution spectra .
the spectropolarimetric data exhibit weak polarization in the continuum , but the line polarization is quite strong ( @xmath8 ) at peak brightness .
|
i.n . and m.l.s .
contributed equally to this work .
the authors gratefully acknowledge w. venstra for useful discussion , a. di michele and p. sassi for raman analysis .
the authors acknowledge financial support from the european commission ( fpvii , grant agreement no : 318287 , landauer ) .
21ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase
10.1021/nl402875 m [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( )
|
the engineering through strain of material properties is very interesting for a wide list of applications , specially for atomically thin membranes made of mos@xmath0 because of its high young modulus and fracture strength .
the modification of the electronic structure and electronic transport under tensile and compressive strain of mos@xmath0 and other transition metal dichalcogenides has been predicted by _
ab initio _ calculations in several works .
a transition from semiconductor to metallic transport has been predicted for single and few layer mos@xmath0 . in this article
we present the observation of this effect on a few layer mos@xmath0 ribbon under a maximum uniaxial tensile strain of 0.14 .
experimental data are accompanied with _
ab initio _ calculations showing that uniaxial strain on few layer mos@xmath0 provokes the closing of the energy band - gap .
molybdenum disulphide , mos@xmath0 , is a layered crystalline solid with an hexagonal structure similar to graphene .
single layer mos@xmath0 is formed by a plane of mo atoms sandwiched and covalently bonded to two planes of s atoms .
few layer and bulk mos@xmath0 is formed by successive stacking of this hexagonal structure through vdw forces . in a bulk form , mos@xmath0 is an indirect semiconductor with energy gap eg= while its single layer counterpart shows a direct transition at the k point with eg=. this material among other transition metal dichalcogenides ( tmds ) has been widely studied in the last years because of their outstanding optical , mechanical and electrical properties along very strong electro - mechanical and opto - mechanical coupling@xcite . in particular
, mos@xmath0 has been suggested to show a transition from semiconductor to metal under mechanical strain@xcite and not only for monolayer mos@xmath0 but also for few - layer mos@xmath0 and even bulk@xcite .
although several studies based on _ ab initio _ calculations have predicted this particular effect no experimental validation has been published until now . in this letter
we report the evidence of the semiconductor - metal transition of mos@xmath0 under tensile strain .
in particular , changes in the direct - current electronic transport properties of mos@xmath0 are characterized for a length mos@xmath0 suspended ribbon under uniaxial tensile strain up to @xmath1=0.14 .
the transition , that occurs at @xmath2 0.12 , is seen as a drastic increment in the current flowing through the device which is related to the closure of the gap that , typically , semiconductors show in their two terminal i - v curve .
first principles band structure calculations of the material are then performed to corroborate the experimental data .
the results shown in this letter paves the way for the development of new micro and nanodevices exploiting the strong electro - mechanical coupling of mos@xmath0 such as strain sensors , straintronic devices , among others .
mos@xmath0 flakes are obtained by mechanical exfoliation from a single crystal @xcite with the `` scotch tape method '' and then transferred to a polydimethylsiloxane ( pdms ) holder as in ref .
@xcite .
an optical microscope is used to locate few layer mos@xmath0 candidates for the analysis . in particular
we focused our attention on the flake reported in fig .
[ f : layers](a ) . to determine the number of layers composing the structure we have performed an optical analysis considering the light transmittance of single - layer mos@xmath0 to be 94.5% @xcite . increasing the number of layer the transmittance decrease linearly proportional to the number of layers , @xmath3 @xcite .
the analysis for the selected flake is reported in fig .
[ f : layers](b ) , where the transmittance profile is relative to the section highlighted in fig .
[ f : layers](a ) .
each horizontal gray line is relative to the expected transmitted light for a specific number of layers starting from the top line corresponding to single layer , @xmath3=1 .
the red curve corresponds to the transmitted light of a 7-layers mos@xmath0 , matching the data obtained for the lighter part of the flake reported in fig .
[ f : layers](a ) .
we also performed raman analysis to further characterize the sample .
mos@xmath0 shows two peaks in the raman spectrum , @xmath4 and @xmath5 , the position and intensity of which have been demonstrated to be indicators of the number of layers @xcite .
fig .
[ f : layers](c ) shows the measured spectrum for the present device ( points ) excited by line and a double gaussian fitted curve ( line ) centered at and which is in good agreement with ref . for @xmath3=7
the relative intensity is @xmath6 , also supporting @xmath3=7 . in order to characterize the semiconductor - to - metal transition of a seven layer mos@xmath0 ribbon ,
we have carried out first principles calculations within density functional theory ( dft ) to investigate the electronic band structure of the material .
we used the siesta package @xcite , norm - conserving pseudopotentials and the perdew - burke - eznerhof parameterization ( pbe ) of the generalized gradient approximation ( gga ) @xcite to obtain the electronic structure of infinite seven layer mos@xmath0 .
an optimized double-@xmath7 polarized basis set to expand the one - electron wave - function is considered . a @xmath8 grid of k - points to sample the brillouin zone is used and the structure is relaxed until all forces are lower than . the unit - cell of stacked mos@xmath0 is shown in fig .
[ f : simulation](a ) .
the values for the energy gap ( eg= ) and the lattice constant ( a= ) are in good agreement with those in the literature@xcite .
strain is introduced to the system by deforming the unit cell along the zig - zag ( x - axis ) and arm - chair ( y - axis ) direction , so @xmath9 .
the evolution of the energy band - gap under tensile strain is shown in fig .
[ f : simulation](c ) decreases as the strain increases .
however , one should note that , while predictions of the pressure coefficients based on dft calculations are very reliable , the band - gaps are notoriously underestimated .
therefore , data shown in fig . [
f : simulation ] ( b ) represent just a trend of @xmath10 . .
a ) atomic structure of stacked mos@xmath0 in the x - y plane , showing bonding between mo and s atoms .
b ) energy gap of seven layer mos@xmath0 as function of applied uniaxial strain along @xmath11 and @xmath12 directions . ] to test the semiconductor to metal transition in our experiment the selected flake has been transferred over a au / cr coated ( ) sio@xmath0 substrate .
the hospital substrate has been mounted on a high precision nanopositioning stage clamping one end of the substrate to the reference frame and the other to the movable part .
the positioning system is capable of a maximum extension of with a pull force of .
thanks to a incision made on the bottom of the substrate the force applied by the stage during elongation is capable of producing a clean crack and split it into two movable parts separated by a gap @xmath13 .
once the substrate is split in two parts the flake is transferred over the trench .
the initial gap was set to @xmath14 determining the length of the metal - semiconductor - metal ( msm ) device to @xmath15 .
after the mos@xmath0 ribbon is transferred , two blobs of epoxy resin mechanically fix both clamps of the ribbon preventing slipping during strain application .
the complete procedure of substrate preparation and transfer of the flake is depicted in detail in fig .
[ f : scheme ] .
flake .
] strain is applied to the msm device by increasing the gap , @xmath13 , being @xmath16 .
the effective value of the gap is directly measured through the integrated positioning sensor of the linear actuator .
a possible way to directly observe semicondutor to metal transition under tensile strain is monitoring the current flowing through the material varying the polarization @xmath17 in a two terminal configuration , _
i.e. _ , the i - v curve
. the typical semiconductor behavior is observed in the suspended unstrained ( @xmath1=0.00 ) mos@xmath0 flake , the i - v curve in this condition is represented by the green curve in fig .
[ f : iv](a ) . in this scenario
the current remains low for @xmath18 . under tensile strain , the current flowing through the flake increases even for @xmath19 , increasing substantially the slope of the curve from to and reaching a linear relation , which corresponds to a metallic behavior ( red curve of fig .
[ f : iv](a ) ) .
the lighter curves presented in fig .
[ f : iv](a ) are relative to intermediate strains between those relative to green and red curves . after increasing the strain beyond @xmath1=0.014
the 7 layer ribbon broke and only the bulk part remained .
the contribution from the bulk mos@xmath0 part of the flake has been then measured and is represented by the dashed line in fig .
[ f : iv](a ) . in this case
no change on the i - v curve was measured as @xmath13 increased .
the i - v curves for unstrained and strained device subtracting the static contribution of the bulk part of the ribbon are presented in fig .
[ f : iv](b ) .
fig .
[ f : iv](c ) show a color map of the current flowing through the device as function of applied strain and polarization voltage , showing an increase in current as function of the voltage for higher applied tensile strains . in fig .
[ f : model](a ) the equivalent circuit of the measured device is presented .
@xmath20 represents the contact resistance , @xmath21 the background impedance that accounts for the bulk mos@xmath0 and parasitic currents . finally , @xmath22 is the strain dependent impedance of the seven layer mos@xmath0 . according to this model
the total conductance of the device can be expressed as : @xmath23 where the conductance @xmath24 is the inverse of the impedance @xmath25 .
we now focus on the conductance of the seven layer mos@xmath0 .
according to ref .
@xcite the conductance of a semiconductor under strain can be expressed as @xmath26 \label{eq:2}\ ] ] where @xmath27 stands for the conductance of the unstrained mos@xmath0 flake , @xmath28 is the boltzmann constant and @xmath29 is the temperature .
the measured conductance of the device is represented by black dots in fig .
[ f : model](b ) .
the fit of eq .
[ eq:1 ] and eq .
[ eq:2 ] , obtained with experimental data is represented by red solid line in fig .
[ f : model](b ) showing a good agreement between experiment and model .
the fitting parameters are reported in table [ t : parameters ] , which are in very good agreement with literature@xcite .
.fitted parameters from the conductance model .
[ cols="^,^",options="header " , ] [ t : parameters ] we have also calculated the gauge factor for the present msm device at @xmath17= ( @xmath30 / \delta \varepsilon $ ] ) to be @xmath31110 in the range 5 - 10% of strain .
the obtained value exceeds the values of conventional and other 2d based strain sensor , like graphene based devices@xcite . = . ] in conclusion , we have measured the semiconductor - metal transition under monoaxial tensile strain of a 7 layer mos@xmath0 ribbon .
an increase of more than an order of magnitude of the current flowing through the device , and a linear i - v relation are obtained at @xmath1=0.014 . a sound model of the conductance as function of the applied strain
is given showing a good agreement between experimental data and theory .
|
the origin of electroweak symmetry breaking and the mechanism that generates the masses of the known fundamental particles is one of the central challenges of particle physics .
the higgs mechanism @xcite in its most general form can be used to explain the observed masses of the @xmath0 and @xmath1 bosons as a consequence of three goldstone bosons ( @xmath2 and @xmath3 ) that end up as the longitudinal components of the massive gauge bosons .
these goldstone bosons are generated by the underlying dynamics responsible for electroweak symmetry breaking .
however , the fundamental nature of this dynamics is still unknown .
two broad classes of electroweak symmetry breaking mechanisms have been pursued theoretically . in one class of theories ,
the electroweak symmetry breaking dynamics is weakly - coupled , while in the second class of theories the dynamics is strongly - coupled @xcite .
the electroweak symmetry breaking dynamics of the standard model ( sm ) employs a self - interacting complex doublet of scalar fields , which consists of four real degrees of freedom @xcite .
elementary scalar dynamics yields a non - zero scalar field vacuum expectation value ( vev ) , which leads to the breaking of the electroweak symmetry .
consequently , three massless goldstone bosons are generated , while the fourth scalar degree of freedom that remains in the physical spectrum is the cp - even neutral higgs boson of the standard model .
precision electroweak data favors a higgs mass below 200 gev , in which case the scalar self - interactions are weak @xcite . in the weakly - coupled approach
to electroweak symmetry breaking , the standard model is very likely embedded in a supersymmetric theory @xcite in order to stabilize the large gap between the electroweak and the planck scales in a natural way @xcite .
the minimal supersymmetric extension of the standard model ( mssm ) employs two complex higgs doublets , resulting in five physical scalar degrees of freedom . in a large range of the mssm parameter space ,
the properties of the lightest scalar of the mssm are nearly indistinguishable from those of the sm higgs boson .
it is possible that the dynamics responsible for electroweak symmetry breaking is strongly - coupled @xcite .
precision electroweak data disfavors the simplest classes of models based on this approach .
nevertheless , it remains possible that the physics of electroweak symmetry breaking is more complicated than suggested by the standard model or its supersymmetric extensions .
however , in this talk i will not consider these theoretical alternatives nor their phenomenological consequences .
in the standard model , the higgs mass is given by : @xmath4 , where @xmath5 is the higgs self - coupling parameter . since @xmath5 is unknown at present , the value of the standard model higgs mass is not predicted .
in contrast , the higgs couplings to fermions [ bosons ] are predicted by the theory to be proportional to the corresponding particle masses [ squared - masses ] . in particular , the sm higgs boson is a cp - even scalar , and its couplings to gauge bosons , higgs bosons and fermions are given by : & & g_h f|f= , g_hvv = , g_hhvv = , [ hsmcouplings1 ] + & & g_hhh = v = , g_hhhh = = , [ hsmcouplings2 ] where @xmath6 or @xmath1 and @xmath7 gev . in higgs production and decay processes , the dominant mechanisms involve the coupling of the higgs boson to the @xmath0 , @xmath1 and/or the third generation quarks and leptons .
note that a @xmath8 coupling ( @xmath9=gluon ) is induced by virtue of a one - loop graph in which the higgs boson couples dominantly to a virtual @xmath10 pair .
likewise , a @xmath11 coupling is generated , although in this case the one - loop graph in which the higgs boson couples to a virtual @xmath12 pair is the dominant contribution .
further details of the sm higgs boson properties are given in .
reviews of the sm higgs properties and its phenomenology , with an emphasis on the impact of loop corrections to the higgs decay rates and cross - sections can be found in refs .
@xcite . for an arbitrary higgs sector ,
the tree - level @xmath13-parameter is given by @xcite 0=1 ( 2t+1)^2 - 3y^2=1 , independently of the higgs vevs , where @xmath14 and @xmath15 specify the weak - isospin and the hypercharge of the higgs representation to which it belongs .
@xmath15 is normalized such that the electric charge of the scalar field is @xmath16 .
the simplest solutions are higgs singlets @xmath17 and hypercharge - one complex higgs doublets @xmath18 .
thus , we shall consider non - minimal higgs sectors consisting of multiple higgs doublets ( and perhaps higgs singlets ) , but no higher higgs representations , to avoid the fine - tuning of higgs vevs .
the two - higgs - doublet model ( 2hdm ) consists of two hypercharge - one scalar doublets . of the eight initial degrees of freedom ,
three correspond to the goldstone bosons and five are physical : a charged higgs pair , @xmath19 and three neutral scalars .
in contrast to the sm where the higgs - sector is cp - conserving , the most general 2hdm allows for higgs - mediated cp - violation ( see for a complete specification of the 2hdm tree - level interactions ) .
if cp is conserved , the higgs spectrum contains two cp - even scalars , @xmath20 and @xmath21 and a cp - odd scalar @xmath22 .
thus , new features of the extended higgs sector include : ( i ) charged higgs bosons ; ( ii ) a cp - odd higgs boson ( if cp is conserved in the higgs sector ) ; ( iii ) possible higgs - mediated cp - violation and neutral higgs states of indefinite cp .
more exotic higgs sectors allow for doubly - charged higgs bosons , etc .
further details of extended higgs sectors can be found in .
the higgs sector of the mssm is a 2hdm , whose yukawa couplings and higgs potential are constrained by supersymmetry ( susy ) . minimizing the higgs potential ,
the neutral components of the higgs fields acquire vevs @xmath23 and @xmath24 , where @xmath25 .
the ratio of the two vevs is an important parameter of the model , @xmath26 .
the five physical higgs particles consist of a charged higgs pair @xmath19 , one cp - odd scalar @xmath22 , and two cp - even scalars @xmath20 , @xmath21 , obtained by diagonalizing the @xmath27 cp - even higgs squared - mass matrix .
all tree - level higgs masses and couplings and the angle @xmath28 ( which parameterizes the diagonalization of the @xmath27 cp - even higgs squared - mass matrix ) can be expressed in terms of two higgs sector parameters , usually chosen to be @xmath29 and @xmath30 .
see refs .
@xcite for further details . at tree level , @xmath31 , which is ruled out by lep data .
but , this inequality receives quantum corrections .
the tree - level higgs mass is shifted due to an incomplete cancellation of particles and their superpartners ( which would be an exact cancellation if supersymmetry were unbroken ) that contribute at one - loop to the higgs self - energy function .
the higgs mass upper bound is then modified to @xcite ^2 ^ 2 + 3g^2 m_t^48 ^ 2 ^ 2 , where @xmath32 governs stop mixing and @xmath33 is the average top - squark squared - mass .
the state - of - the - art computation includes the full one - loop result , all the significant two - loop contributions , some of the leading three - loop terms , and renormalization - group improvements @xcite .
the final conclusion is that @xmath34 gev [ assuming that the top - squark mass is no heavier than about 2 tev ] .
this upper bound is reached when @xmath35 and @xmath36 in the so - called _ maximal mixing scenario _ , which corresponds to choosing @xmath37 so that the predicted value of @xmath38 is maximal .
tree - level couplings of the mssm higgs bosons with gauge bosons are often suppressed by an angle factor , either @xmath39 or @xmath40 , as shown in the table below .
@xmath41 \noalign{\vskip3pt } \hh w^+w^- & \hl w^+w^- & \qquad{\hbox{\rm{\phantom{a}---\phantom{a}}}}\\ \hh zz & \hl zz & \qquad{\hbox{\rm{\phantom{a}---\phantom{a}}}}\\ z\ha\hl & z\ha\hh & zh^+h^-\,,\,\,\gamma h^+h^-\\ w^\pm h^\mp\hl & w^\pm h^\mp\hh & w^\pm h^\mp\ha \\ \end{array}\ ] ] tree - level higgs - fermion couplings may be either suppressed or enhanced with respect to the sm value , @xmath42 . the charged higgs boson couplings to fermion pairs , with all particles pointing into the vertex , are : g_h^- t|b= & & , + g_h^- ^+ = & & , where @xmath43 , and the neutral higgs boson couplings are : b|b ( or ^+ ^- ) : & & - = ( - ) -(- ) , + t|t : & & = ( - ) + ( - ) , + b|b ( or ^+ ^- ) : & & = ( - ) + ( - ) , + t|t : & & = ( - ) -(- ) , + b |b ( or ^+ ^- ) : & & _ 5 , + t |t : & & _ 5 , where the @xmath44 indicates a pseudoscalar coupling . especially noteworthy is the possible @xmath30-enhancement of certain higgs - fermion couplings .
typically , @xmath45 in most mssm models . in many models with extended higgs sectors
, a parameter regime exists in which one higgs boson is light ( of order @xmath46 ) and all other higgs scalars are very heavy @xmath47 ) . in this case , one can formally integrate out the heavy scalar states .
the effective low - energy higgs theory is precisely that of the sm higgs boson .
this is called the decoupling limit @xcite .
the mssm exhibits the decoupling limit for @xmath48 . in this case , it is easy to verify that @xmath49 , up to corrections of @xmath50 , and @xmath51 up to corrections of @xmath52 .
examining the higgs couplings listed above , one can check that in the limit of @xmath53 , all @xmath20 couplings to sm particles approach their sm limits , as expected .
from 19892000 , experiments at lep searched for @xmath54 ( where one of the @xmath1-bosons is on - shell and one is off - shell ) .
no significant evidence was found leading to a lower bound on the higgs mass , @xmath55 at @xmath56 @xcite .
recent data from the tevatron extends the disallowed higgs mass region by excluding @xmath57 at @xmath56 @xcite
. the mssm higgs mass bounds are more complicated , since they depend on a variety of mssm parameters . since supersymmetric radiative corrections to the lightest mssm higgs mass must be significant to avoid exclusion , the mssm higgs mass bounds depend sensitively on the multi - dimensional mssm parameter space . in the maximal - mixing scenario ,
the lep higgs search rules out ( at 95% cl ) charged higgs bosons with @xmath58 gev and neutral higgs masses with @xmath59 gev and @xmath60 gev @xcite . in certain other scenarios ,
it is possible to significantly relax these bounds .
for example , in the cpx scenario @xcite , supersymmetric radiative corrections introduce cp - violating phenomena into the mssm higgs sector , in which case an arbitrarily light higgs boson is still allowed for @xmath61 @xcite .
precision electroweak data puts indirect bounds on the higgs mass .
one can fit a plethora of electroweak observables , based on the predictions of the sm , in which the higgs mass is allowed to float .
the global fit of precision electroweak data yields a @xmath62 distribution for the goodness of fit as a function of the higgs mass . if the direct higgs search bounds are omitted , the favored central value of the higgs mass obtained by the lep electroweak working group is 89 gev , with a one - sigma experimental uncertainty of @xmath63 and @xmath64 gev @xcite .
the corresponding one - sided 95% cl upper bound excludes @xmath65 gev . including the lep higgs search data , the upper bound increases to 185 gev .
similar results are obtained by the ` gfitter ` collaboration , which quotes @xcite : m_h=120.6^+17.0[+34.3]_-5.2[-6.2 ] gev , based on direct and indirect data , where both 1@xmath66 and 2@xmath66 error bars are given ( the latter in the square brackets ) . the minimum @xmath62 associated with the ` gfitter ` fit is @xmath67 for 14 degrees of freedom .
if new physics beyond the standard model ( sm ) exists , then one expects shifts in the @xmath68 and @xmath1 masses due to new contributions to the one - loop corrections . in many cases
, these effects can be parameterized in terms of two quantities , @xmath69 and @xmath14 introduced by peskin and takeuchi @xcite : t & & - , + s & & - ( ) - , where @xmath70 , @xmath71 , and barred quantities are defined in the @xmath72 scheme evaluated at @xmath46 . the @xmath73 are the new physics contributions to the one - loop gauge boson vacuum polarization functions .
the region in the @xmath69@xmath14 plane consistent with precision electroweak data is shown in fig .
[ st](a ) . vs. @xmath14 plane based on the ` gfitter ` global standard model fit to electroweak precision data .
the point @xmath74 corresponds to @xmath75 gev and @xmath76 gev .
( b ) in the right panel , the ` gfitter ` fit to electroweak precision data is exhibited for the littlest higgs model of ( with @xmath14-parity ) for the parameters shown .
heavier higgs masses are allowed , if the effects of the higgs bosons to the gauge boson vacuum polarization functions are compensated by the effects due to new particles and interactions contained in the littlest higgs model .
taken from .
[ st],title="fig:",width=298 ] vs. @xmath14 plane based on the ` gfitter ` global standard model fit to electroweak precision data .
the point @xmath74 corresponds to @xmath75 gev and @xmath76 gev .
( b ) in the right panel , the ` gfitter ` fit to electroweak precision data is exhibited for the littlest higgs model of ( with @xmath14-parity ) for the parameters shown .
heavier higgs masses are allowed , if the effects of the higgs bosons to the gauge boson vacuum polarization functions are compensated by the effects due to new particles and interactions contained in the littlest higgs model .
taken from .
[ st],title="fig:",width=298 ] in order to avoid the conclusion of a light higgs boson , new physics beyond the sm must enter at an energy scale between 100 gev and 1 tev .
this new physics can be detected at the lhc either through direct observation of new physics beyond the sm or by improved precision measurements at future colliders that can detect small deviations from sm predictions . although the precision electroweak data is suggestive of a light higgs boson , one can not definitively rule out a significantly heavier higgs boson if the new physics conspires to yield a total contribution to @xmath69 and @xmath14 that is consistent with the precision electroweak data [ e.g. , see fig .
[ st](b ) ] .
the tevatron will continue to take data through the end of 2011 .
in addition to an increased integrated luminosity , there is still room for some improvements in the higgs search analysis .
the possibility of extending the tevatron run by another three years is currently under discussion . to facilitate this discussion , a collaboration of fermilab theorists and experimentalists produced a white paper making the case for an extended run iii of the tevatron from the end of 2011 through 2014 @xcite .
the projected improvement of the tevatron higgs search is shown in fig . [ tevatron ] .
in table 1 , the expected sensitivity of the tevatron higgs search for higgs masses of 115 gev , 130 gev and 145 gev as a function of the analyzable luminosity per experiment is given .
.sensitivity to the standard model higgs boson combining all modes .
the low mass @xmath77 gev mode is principally @xmath78 ; the higher mass @xmath79 gev mode is principally @xmath80 . taken from . [
cols="^,^,^,^",options="header " , ] after the higgs boson discovery , one must check that its properties are consistent with the theoretical expectations .
measurements of the spin , parity and couplings to gauge bosons and fermions typically need a large data sample ( which requires tens to hundreds of fb@xmath81 )
. however , in some special circumstances , one can discriminate between different hypotheses for spin and parity with a small data sample .
for example , for higgs boson masses above about 130 gev , it will be possible to isolate a data sample of @xmath82 . by analyzing the distributions and correlations of the five relevant angular variables that describe the @xmath83 events
, one can achieve a median expected discrimination significance of @xmath84 with as few as 19 events for @xmath85 gev and even better discrimination for the off - shell decays of an @xmath86 gev higgs boson @xcite , as shown in table 2 .
thus , if a higgs boson is discovered at the lhc with @xmath87 , then it may already be possible to confirm a preference for the expected @xmath88 spin - parity assignment with a small sample of higgs events . in the decoupling limit ,
the properties of the lightest cp - even higgs boson of the mssm ( denoted by @xmath20 ) are nearly identical to that of the sm higgs boson .
thus , all the sm higgs search techniques apply without modification in the mssm .
moreover , the other mssm higgs bosons , @xmath19 , @xmath89 and @xmath90 , which are significantly heavier than @xmath91 ( and @xmath46 ) , are roughly degenerate in mass ( cf .
section 2.4 ) . in the opposite limit
( far from the decoupling limit ) , all physical mssm higgs bosons are typically below 200 gev in mass and no single higgs state possesses couplings that precisely match those of the sm higgs boson @xcite .
cross - sections and branching ratios for the mssm higgs boson ( which depend on @xmath92 as well as the corresponding higgs mass ) can be found in refs .
@xcite . here
, we simply note that : * gluon - gluon fusion can produce both neutral cp - even and cp - odd higgs bosons .
* @xmath93 fusion ( @xmath6 or @xmath1 ) can produce only cp - even higgs bosons ( at tree - level ) .
moreover , in the decoupling limit , the heavy cp - even higgs boson is nearly decoupled from the @xmath93 channel . *
neutral higgs bosons can be produced in association with @xmath94 and with @xmath10 in gluon - gluon scattering . *
charged higgs bosons can be produced in association with @xmath95 in gluon - gluon scattering . *
if @xmath96 , then @xmath97 is an allowed decay , and the dominant @xmath19 production mechanism is via @xmath10 production .
* higgs bosons can be produced in pairs ( e.g. , @xmath98 , @xmath99 , @xmath100 ) . *
higgs bosons can be produced in cascade decays of susy particles . *
the total width of any one of the mssm higgs bosons is always @xmath101 of its mass [ cf .
* higgs search strategies depend on the location of the model in the @xmath29@xmath30 plane .
simulations by the atlas and cms collaborations shown in fig .
[ wedge ] exhibit which regions of the @xmath29@xmath92 plane are sensitive to the mssm higgs search . in particular , for large values of @xmath102 gev ( corresponding to the decoupling limit ) , there is no sensitivity for the discovery of the heavy mssm higgs states , @xmath19 , @xmath89 and @xmath90 , for moderate values of @xmath92 .
this is the infamous `` lhc wedge '' region , in which the @xmath103 of the mssm ( whose properties are nearly indistinguishable from the sm higgs boson ) is the only mssm higgs boson that can be discovered at the lhc .
further details can be found in refs . @xcite .
i would like to thank jose valle for providing me with the opportunity to visit valencia and to speak at the pascos-2010 conference .
i am especially grateful to jose and his colleagues at the universitat de valncia for their kind hospitality during my visit .
i also appreciate a number of useful comments and suggestions from sebastian grab and lorenzo ubaldi , who read the draft version of this manuscript . this work is supported in part by the u.s .
department of energy , under grant number de - fg02 - 04er41268 .
99 p.w . higgs , phys . rev
* 12 * ( 1964 ) 132 ; phys . rev .
* 145 * ( 1966 ) 1156 ; f. englert and r. brout , phys . rev
* 13 * ( 1964 ) 321 ; g.s .
guralnik , c.r .
hagen , and t.w.b .
kibble , phys .
* 13 * ( 1964 ) 585 .
gunion , h.e .
haber , g.l .
kane and s. dawson , _ the higgs hunter s guide _
( westview press , boulder , co , 2000 ) .
the aleph , cdf , d0 , delphi , l3 , opal , sld collaborations , the lep electroweak working group , the tevatron electroweak working group , and the sld electroweak and heavy flavor groups , cern - ph - ep/2009 - 023 ( 13 november , 2009 ) ; the most recent updated results can be found at ` http://lepewwg.web.cern.ch/lepewwg/ ` .
nilles , phys .
* 110 * ( 1984 ) 1 ; h.e .
haber and g.l .
kane , phys .
* 117 * ( 1985 ) 75 ; s.p .
martin , `` a supersymmetry primer , '' in _
perspectives on supersymmetry ii _ , edited by g.l .
kane ( world scientific , singapore , 2010 ) pp .
1153 [ arxiv : hep - ph/9709356v5 ] .
e. witten , nucl .
b * 188 * ( 1981 ) 513 ; l. susskind , phys .
* 104 * ( 1984 ) 181 . for a recent review , see e.g. , c.t . hill , e. h. simmons , phys .
* 381 * ( 2003 ) 235 .
[ hep - ph/0203079 ] .
kniehl , phys .
* 240 * ( 1994 ) 211 .
m. carena and h.e .
haber , prog .
* 50 * ( 2003 ) 63 [ arxiv : hep - ph/0208209 ] .
a. djouadi , phys .
* 457 * ( 2008 ) 1 [ arxiv : hep - ph/0503172 ] .
tsao , in _ proceedings of the 1980 guangzhou conference on theoretical particle physics _ , edited by h. ning and t. hung - yuan ( science press , beijing , 1980 ) p. 1240 .
haber and d. oneil , phys .
d * 74 * ( 2006 ) 015018 [ arxiv : hep - ph/0602242 ] ; erratum ibid .
* 74 * ( 2006 ) 059905 .
a. djouadi , phys .
* 459 * ( 2008 ) 1 [ arxiv : hep - ph/0503173 ] .
h.e . haber and r. hempfling ,
* 66 * ( 1991 ) 1815 ; y. okada , m. yamaguchi and t. yanagida , prog .
. phys .
* 85 * ( 1991 ) 1 ; j.r .
ellis , g. ridolfi and f. zwirner , phys .
b * 257 * ( 1991 ) 83 .
see e.g. , g. degrassi , s. heinemeyer , w. hollik , p. slavich and g. weiglein , eur .
j. c * 28 * ( 2003 ) 133 [ arxiv : hep - ph/0212020 ] ; b.c .
allanach , a. djouadi , j.l .
kneur , w. porod and p. slavich , jhep * 0409 * ( 2004 ) 044 [ arxiv : hep - ph/0406166 ] ; s.p .
martin , phys .
d * 75 * ( 2007 ) 055005 [ arxiv : hep - ph/0701051 ] ; p. kant , r.v .
harlander , l. mihaila and m. steinhauser , jhep * 1008 * ( 2010 ) 104 [ arxiv:1005.5709 [ hep - ph ] ] .
h.e . haber and y. nir ,
b * 335 * ( 1990 ) 363 ; j.f . gunion and h.e .
haber , phys .
d * 67 * ( 2003 ) 075019 [ arxiv : hep - ph/0207010 ] .
r. barate _
[ lep working group for higgs boson searches and aleph and delphi and l3 and opal collaborations ] , phys .
lett . b * 565 * ( 2003 ) 61 .
[ hep - ex/0306033 ] .
the tevnph working group for the cdf and d0 collaborations , fermilab - conf-10 - 257-e ( july 28 , 2010 ) [ arxiv:1007.4587 [ hep - ex ] ] .
the aleph , delphi , l3 and opal collaborations , and the lep higgs working group , lhwg note 2001 - 04 ( july 9 , 2001 ) ; lhwg note 2001 - 05 ( july 4 , 2001 ) .
updates can be found at : ` http://lephiggs.web.cern.ch/lephiggs/www/welcome.html ` .
demir , phys .
d * 60 * ( 1999 ) 095007 [ arxiv : hep - ph/9905571 ] ; a. pilaftsis and c.e.m .
wagner , nucl .
b * 553 * ( 1999 ) 3 [ arxiv : hep - ph/9902371 ] ; s.y .
choi , m. drees and j.s .
lee , phys .
b * 481 * ( 2000 ) 57 [ arxiv : hep - ph/0002287 ] ; m. carena , j.r .
ellis , a. pilaftsis and c.e.m .
wagner , nucl .
b * 586 * ( 2000 ) 92 [ arxiv : hep - ph/0003180 ] ; s. heinemeyer , eur .
j. c * 22 * ( 2001 ) 521 [ arxiv : hep - ph/0108059 ] ; t. ibrahim and p. nath , phys .
d * 63 * ( 2001 ) 035009 [ arxiv : hep - ph/0008237 ] ; ibid .
* 66 * ( 2002 ) 015005 [ arxiv : hep - ph/0204092 ] .
m.e . peskin and t. takeuchi ,
lett . * 65 * ( 1990 ) 964 ; phys .
d * 46 * ( 1992 ) 381 .
n. arkani - hamed , a.g .
cohen , e. katz and a.e .
nelson , jhep * 0207 * ( 2002 ) 034 [ arxiv : hep - ph/0206021 ] ; h.c . cheng and i.
low , jhep * 0408 * ( 2004 ) 061 [ arxiv : hep - ph/0405243 ] ; j. hubisz , p. meade , a. noble and m. perelstein , jhep * 0601 * ( 2006 ) 135 [ arxiv : hep - ph/0506042 ] .
m. carena _ et al .
_ , `` run iii : continued running of the tevatron collider beyond 2011 '' ( may 26 , 2010 ) , available from ` http://beamdocs.fnal.gov/ad-public/docdb/showdocument?docid=3617 ` . u. aglietti _ et al . _
[ tev4lhc higgs working group ] , `` tevatron - for - lhc report : higgs , '' fermilab - conf-06 - 467-e - t [ arxiv : hep - ph/0612172 ] .
g. aad _ et al . _
[ the atlas collaboration ] , _ expected performance of the atlas experiment : detector , trigger and physics _ , cern - open-2008 - 020 ( december , 2008 ) [ arxiv:0901.0512 [ hep - ex ] ] .
a. de rujula , j. lykken , m. pierini , c. rogan and m. spiropulu , phys .
d * 82 * ( 2010 ) 013003 [ arxiv:1001.5300 [ hep - ph ] ] .
e. boos , a. djouadi , m. muhlleitner and a. vologdin , phys .
d * 66 * ( 2002 ) 055004 [ arxiv : hep - ph/0205160 ] ; e. boos , a. djouadi and a. nikitenko , phys . lett .
b * 578 * ( 2004 ) 384 [ arxiv : hep - ph/0307079 ] .
s. abdullin _ et al .
_ , eur .
j. c * 39 * , s2 ( 2005 ) s41-s61 .
|
discovering the higgs boson is one of the primary goals of both the tevatron and the large hadron collider ( lhc ) .
the present status of the higgs search is reviewed and future prospects for discovery at the tevatron and lhc are considered .
this talk focuses primarily on the higgs boson of the standard model and its minimal supersymmetric extension .
theoretical expectations for the higgs boson and its phenomenological consequences are examined .
scipp-10/17 + november , 2010 +
|
the three - dimensional distribution of galaxies observed in redshift surveys differs from the true one since the distance to each galaxy can not be determined by its redshift @xmath8 only ; for @xmath9 the peculiar velocity of galaxies , typically @xmath10 , contaminates the true recession velocity of the hubble flow ( e.g. , davis & peebles 1983 ; kaiser 1987 ; hamilton 1992 ) , while the true distance for objects at @xmath11 sensitively depends on the ( unknown and thus assumed ) cosmological parameters .
this hampers the effort to understand the true distribution of large - scale structure of the universe .
nevertheless such redshift - space distortion effects are quite useful since through the detailed theoretical modeling , one can derive the peculiar velocity dispersions of galaxies as a function of separation ( davis & peebles 1983 ) , and also can infer the cosmological density parameter @xmath12 and the dimensionless cosmological constant @xmath13 , for instance ( ballinger , peacock & heavens 1996 ; matsubara & suto 1996 ; suto & matsubara 1996 ) . theoretically , a conventional tool to predict small - scale relative pairwise peculiar velocity dispersions is the cosmic virial theorem ( peebles 1976 ; hereafter cvt ) . in its simplest form , the prediction is based on the idealistic assumption that galaxies are treated as point particles . as peebles ( 1976 ) remarked , however , the finite size effect is very significant on scale @xmath14 even if one considers a radius of @xmath15 for typical galactic halos . in this paper
we explore this realistic effect in more details and discuss its important impact on the application of the cvt to real observations .
we will also give reliable predictions for the small - scale velocity dispersions of galaxies in cobe normalized cdm models .
the plan of the paper is as follows ; section 2 briefly describes the cvt both in its simplest and in its more realistic form .
the finite size effect is considered in 3 , first based on a perturbation theory ( 3.1 ) and then on a full numerical analysis ( 3.2 ) . in
4 we predict velocity dispersion of galaxies on small scales for cobe normalized cdm models .
discussion and implications of our result , and our main conclusions are given in 5 .
assuming that galaxy pairs are gravitationally in `` statistical equilibrium '' on small scales , peebles ( 1976 , 1980 ) derived an expression for their relative ( one - dimensional ) peculiar velocity dispersion as a function of their separation @xmath16 : @xmath17 where @xmath18 is the mean density of the universe , and @xmath19 and @xmath20 are the two- and three - point correlation functions of _
mass_. the observed two- and three - point correlation functions of _ galaxies _ , @xmath21 and @xmath22 ,
are well approximated by the following forms ( groth & peebles 1977 ; davis & peebles 1983 ) : @xmath23\end{aligned}\ ] ] where @xmath24 , @xmath25 , @xmath26 ( @xmath27 is the dimensionless hubble constant @xmath28 in units of @xmath29 ) .
thus it is reasonable to assume that the two- and three - point correlation functions of _ mass _ , @xmath30 and @xmath31 , also obey the same scaling except for the overall amplitudes : @xmath32 then a straightforward computation yields @xmath33 where @xmath34 \right\ } .
\label{eq : i0gamma}\end{aligned}\ ] ] numerically @xmath35 , @xmath36 , and @xmath37 .
more specifically the cvt predicts the small - scale peculiar velocity dispersion as ( peebles 1976 ; suto 1993 ) @xmath38 the above simple estimate can be improved more realistically by including the finite size of the galaxies , which suppresses the effective gravitational force between pairs , and also possibly changes the behavior of the two - point correlation functions on small scales .
in fact peebles ( 1976 ) already remarked that the above two effects may change the predictions of the velocity dispersions significantly .
more recently bartlett & blanchard ( 1996 ) also discussed the importance of the local mass distribution around pairs in the cvt .
our principal aim here is to describe these effects more extensively using both the perturbation analysis and the full numerical integration . for this purpose
we take account of the former effect by softening the gravitational force according to the plummer law with the softening radius @xmath2
. then equation ( [ eq : cvt1 ] ) is modified as @xmath39 we model the latter effect by incorporating a nonzero core radius @xmath0 as follows : @xmath40 the two modifications can be still described in equation ( [ eq : cvt1d ] ) simply by replacing @xmath41 with @xmath42 \nonumber \\ & & = 2\pi ( 1+y)^\gamma \int_0^\infty { t^3 dt \over ( t^2+w^2)^{3/2 } } \int_1^\infty ds \left [ { 1 \over ( s+y)^\gamma } + { 1 \over ( t+y)^\gamma}\right ] k(s , t , y;\gamma ) , \label{eq : iyw}\end{aligned}\ ] ] where @xmath43 , @xmath44 , and @xmath45 \nonumber \\ & + & { s^2+t^2 - 3y^3 \over 2s^2t^2 ( 2-\gamma ) } \left[(s+t+y)^{2-\gamma } - ( |s - t|+y)^{2-\gamma}\right ] \nonumber \\ & + & { y(y^2-s^2-t^2 ) \over 2s^2t^2 ( 1-\gamma ) } \left[(s+t+y)^{1-\gamma } - ( |s - t|+y)^{1-\gamma}\right ] .\end{aligned}\ ] ] the integral @xmath46 is divergent for @xmath5 if @xmath47 .
nonzero values for @xmath48 and/or @xmath49 make @xmath46 finite for @xmath50 , therefore incorporating the finite size of galaxies in this way widens the applicability of the cvt compared with its simplest form ( [ eq : cvt1d ] ) .
for realistic cdm models , the mass two - point correlation functions on small scales may be better described by equation ( [ eq : xirc ] ) with @xmath51 and @xmath52 , as will be seen below .
in the case of @xmath4 , the integral @xmath46 converges even for @xmath47 .
then one may treat the effect of nonzero @xmath48 and @xmath49 via perturbation theory . consider the case of @xmath53 first .
let us set @xmath54 ( @xmath55 is a parameter greater than unity and will be determined below ) , and try to find the taylor expansion with respect to @xmath56 : @xmath57 the derivative @xmath58 is explicitly given as @xmath59 k(s , t , u^\alpha;\gamma ) \nonumber\\ & - & \alpha\gamma u^{\alpha-1 } ( 1+u^\alpha)^\gamma \int_0^\infty dt \int_1^\infty ds \left [ { 1 \over ( s+u^\alpha)^{\gamma+1 } } + { 1 \over ( t+u^\alpha)^{\gamma+1}}\right ] k(s , t , u^\alpha;\gamma ) \nonumber \\ & - & \alpha\gamma u^{\alpha-1 } ( 1+u^\alpha)^\gamma \int_0^\infty dt \int_1^\infty ds \left [ { 1 \over ( s+u^\alpha)^{\gamma } } + { 1 \over ( t+u^\alpha)^{\gamma}}\right ] k(s , t , u^\alpha;\gamma+1 ) .\end{aligned}\ ] ] in the limit of @xmath60 , only the second term gives a non - vanishing contribution : @xmath61 .
\label{eq : iulim}\end{aligned}\ ] ] by setting @xmath62 and then considering the limit @xmath63 , one finds that only the choice of @xmath64 gives the finite value for equation ( [ eq : iulim ] ) which is contributed only by the term proportional to @xmath65 in the square bracket : @xmath66 the above method can be also applied to the perturbative analysis for @xmath67 and @xmath68 , and a straightforward calculation yields @xmath69 where @xmath70 is the beta function .
the above expressions clearly show the importance of the finite size effect for the observational value @xmath71 .
since the lowest - order correction term begins with @xmath72 , even a very small value of @xmath48 changes the value of @xmath46 significantly , which will be discussed in details below .
figure [ fig : iywg18 ] plots the results of the numerical evaluation of equation ( [ eq : iyw ] ) for @xmath71 in comparison with the perturbative expansions ( [ eq : iy0 ] ) to ( [ eq : iyy ] ) . in practice ,
here and in what follows we set @xmath73 and @xmath74 , respectively , when we numerically compute the values of @xmath75 and @xmath67 .
the perturbation result is in excellent agreement with the numerical evaluation up to @xmath76 for @xmath77 and @xmath78 , and up to @xmath79 for @xmath80 .
also the perturbation correction turns out to provide a reasonably good approximation even up to @xmath81 .
thus for astrophysically relevant values of @xmath2 and @xmath0 , the perturbation formulae ( eqs.[[eq : iy0 ] ] to [ [ eq : iyy ] ] ) are very useful if @xmath82 .
it should be noted that even very small values of @xmath48 and/or @xmath49 significantly decrease @xmath83 relative to the idealistic point - mass approximation @xmath84 ; for @xmath71 , @xmath85 is about 0.8 even if @xmath48 or @xmath49 is @xmath86 . on the other hand ,
the dependence on @xmath48 and @xmath49 is fairly weak as indicated by the perturbation correction ( [ eq : iy0 ] ) to ( [ eq : iyy ] ) .
therefore in a sense most of the effect might be interpreted as the inherent offset originated from the point - mass approximation itself .
if @xmath5 , on the other hand , the above perturbative approach does not work and one should rely on the fully numerical integration .
figures [ fig : iygamma ] and [ fig : iyygamma ] show such results ; @xmath75 and @xmath87 for various values of @xmath88 .
we find the following accurate fitting formulae for @xmath89 : @xmath90^{-1 } , \\
i(y , y;\gamma ) & = & b [ ( y / y_b)^{\alpha_b } + ( y / y_b)^{\beta_b } ] ^{-1 } , \end{aligned}\ ] ] where @xmath91 at least in the range of @xmath92 , the agreement is excellent ( figs . [ fig : iygamma ] and [ fig : iyygamma ] ) between the numerical results and the fitting formulae .
let us consider the predictions for the small - scale velocity dispersions on the basis of the cvt with the finite size effect . with nonzero @xmath2 and @xmath0 , equation ( [ eq : cvt1d ] )
should be explicitly rewritten as @xmath93 due to the @xmath16-dependence of @xmath94 , @xmath95 is not described by a power - law in general , but is still proportional to @xmath96 .
figure [ fig : v12g18 ] plots @xmath97 for @xmath71 . as discussed in 3 , even fairly small values for @xmath2 and @xmath0 can reduce the predictions appreciably relative to the case of the point - mass approximation ( @xmath6 ) .
furthermore , when @xmath52 , the softening of the force has negligible effect if @xmath98 . since the observed two - point correlation functions of galaxies do not show any noticeable departure from the single power - law at least for @xmath99 , the choice of @xmath100 plotted here might be an extreme example , and already overestimate the finite size effect . similarly the results for various @xmath88 are plotted in figure [ fig : v12gamma ] for @xmath100 ( dashed curves ) and @xmath101 ( solid curves ) .
the results for larger @xmath88 become more sensitive to the values of @xmath2 and @xmath0 . given the fact that the point - mass approximation fails for @xmath50 , this is reasonable .
it should be also pointed out here that even when @xmath5 @xmath102 increases as @xmath16 because @xmath94 increases more strongly than @xmath103 ; in the case of @xmath100 and @xmath104 relevant for figure [ fig : v12gamma ] , @xmath105 is proportional to @xmath106 = @xmath107 , @xmath108 , @xmath109 , and @xmath110 for @xmath111 , @xmath112 , @xmath113 , and @xmath114 , respectively .
when @xmath115 , @xmath116 , @xmath102 approaches @xmath117 and thus significantly levels off ( around @xmath118 for @xmath100 and @xmath119 for @xmath101 in fig.[fig : v12cdmr ] ) . in any case
, this implies that the scale - dependence of @xmath105 is fairly _ insensitive _ to the value of @xmath88 .
the results presented above are fairly general in the sense that they should be valid as long as the cvt is correct and equations ( [ eq : hier ] ) , ( [ eq : cvtrs ] ) and ( [ eq : xirc ] ) are reasonably good approximation to the reality .
nevertheless the velocity dispersion @xmath102 is still crucially dependent on the parameters including @xmath12 , @xmath120 , @xmath121 , @xmath122 , @xmath88 as well as @xmath2 and @xmath0 which we have discussed in details here . therefore to proceed further one has to assume some sort of specific models . for this purpose
cold dark matter ( cdm ) models would provide good working hypotheses . in particular very accurate semi - analytical fitting formulae for @xmath123 have been worked out in the literature , which have been successfully tested against high - resolution n - body simulations ( hamilton et al .
1991 ; jain , mo & white 1995 ; peacock & dodds 1994 , 1996 ; mo , jing & brner 1996 ) .
cdm models with the primordial scale - invariant ( harrison -zeldovich ) spectrum have the power spectrum @xmath124 which is asymptotically proportional to @xmath125 for large @xmath126 .
this implies that the two - point correlation function in these models should approach constant on the corresponding scales ; for a purely scale - free primordial spectra of @xmath127 , @xmath123 is proportional to @xmath128 in a linear regime and to @xmath129 in a stable clustering ( strongly nonlinear ) regime .
in other words , we naturally expect the small - scale flattening of the form ( [ eq : xirc ] ) even independently of the finite size effect of galaxies .
this is another reason why the present analysis is very important in realistic cosmological scenarios .
given the cosmological parameters @xmath12 , @xmath13 , and @xmath27 , the normalization constant of the fluctuation spectrum can be determined by cobe data ( e.g. , sugiyama 1995 ) , and thus the ( non - linear ) mass correlation function @xmath123 can be analytically calculated ( peacock & dodds 1996 ; mo et al .
we then determine @xmath122 , @xmath88 and @xmath0 by fitting the calculated @xmath123 to equation ( [ eq : xirc ] ) .
so determined @xmath0 does not represent the finite size effect of galaxies , and if it is below @xmath130 , for instance , we may set @xmath101 to include the effect .
therefore as long as the hierarchical relation ( [ eq : hier ] ) holds for the two- and three - point correlation functions of mass , the only remaining parameter is @xmath120 .
although recent numerical and semi - analytical studies indicate that @xmath120 is weakly dependent on the scale ( suto 1993 ; matsubara & suto 1994 ; suto & matsubara 1994 ; jing & brner 1996 ; mo et al .
1996 ) , it is about @xmath131 on scales less than @xmath132 .
we use the following expression for the cobe 2yr data normalization @xmath133 by nakamura ( 1996 ) who performed a fit to the numerical results by sugiyama ( 1995 ) : @xmath134 where @xmath135 } , \\
f(x ) & \equiv & ( 18.1x^{0.2}+7.17x^{0.5}+22.1x+1.23x^2)^{-1 } .\end{aligned}\ ] ] for definiteness we adopt the baryon density parameter @xmath136 .
note that the cobe 2yr data give a slightly larger normalization amplitude ( roughly by 10 percent ) compared with the cobe 4yr data ( sugiyama 1996 , in private communication ) .
our conclusions below , however , are insensitive to this issue .
table 1 : fitted parameters for cdm models [ cols=">,>,>,^,^,^,^",options="header " , ] figure [ fig : xifit ] plots the two - point correlation functions of mass based on the peacock & dodds ( 1996 ) model and our fits to the form ( [ eq : xirc ] ) for the cdm models in table 1 .
the parameter fitting is performed over the range @xmath137 .
although the fit does not seem so good especially on small scales , any single power - law form ( i.e. , @xmath138 ) yields much poorer fits and the inclusion of the nonzero @xmath0 definitely improve the theoretical prediction .
figure [ fig : v12cdmr ] shows @xmath139 for the three cdm models in table 1 .
the model parameters except for @xmath120 are completely determined from theory and the fitting , and therefore the result is fairly definite .
as mentioned above , recent numerical and semi - analytical results suggest @xmath140 on scales less than @xmath132 . with this value in mind ,
figure [ fig : v12cdmr ] implies that @xmath141 cdm models disagree with the observed values @xmath142 at @xmath143 ( davis & peebles 1983 ; mo , jing & brner 1993 ; fisher et al .
1994 ; marzke et al .
1995 ) by a factor of @xmath144 .
although the conflict with the @xmath141 model has been pointed out frequently in the literature ( e.g. , peebles 1976 , 1980 ; davis & peebles 1983 ; suto 1993 ) , our present prediction is the most precise one from a theoretical point of view . finally in figure [
fig : v12omega ] , we plot @xmath145 as a function of @xmath12 both for @xmath146 and @xmath147 models using the same procedure described here .
clearly only the low - density models are compatible with the observations even if one allows for a factor of @xmath148 uncertainty of the sample - to - sample variation ( suto 1996 ; suto & jing 1996 ) .
it is interesting to note from the figure that softening in the force with @xmath149 has almost negligible effect on the prediction of @xmath150 in the three cdm models ; this is simply because these models have either @xmath151 or @xmath152 .
we have presented a realistic refinement of the cosmic virial theorem by taking into account of the finite size effect of galaxies .
the cvt in its conventional form which treats galaxies as point particles appreciably overestimates small - scale velocity dispersions of galaxies .
although this is already remarked by peebles ( 1976 ) and in a slightly different context by bartlett & blanchard ( 1996 ) , we presented more detailed analysis on the basis of a perturbation theory and systematic numerical calculations .
in particular the former provides a clear physical insight into the finite size effect .
our perturbation formulae ( eqs.[[eq : iy0 ] ] [ [ eq : iyy ] ] ) , which are accurate enough [ better than 20% in accuracy for @xmath46 ] for most purposes , can be easily and reliably used to correct for the flattening of mass correlation @xmath154 and the force softening for @xmath4 .
we have also presented accurate fitting formulae for @xmath46 for the case of @xmath5 .
moreover , we have predicted the pairwise velocity dispersion for cdm models with the cobe fluctuation normalization . as our result indicates that the flattening of mass correlation @xmath154 has significant effect on the cvt
, one should be very cautious to infer @xmath12 from observational measurements of galaxy - galaxy correlation function @xmath155 and galaxy pairwise velocity dispersions .
although current observations of galaxy clustering have not revealed any noticeable flattening in @xmath155 on small scales , non - linear biasing mechanisms may have complicated the relation between @xmath155 and @xmath154 . moreover , the statistical significance of galaxy clustering on very small scales @xmath156 is still rather poor .
in fact , intrinsic flattening of correlation functions on small scales is expected generally for cdm models with a harrison -zeldovich primordial spectrum .
we have quantitatively pointed out the importance of taking into account such flattening in @xmath154 when one applies the cvt , otherwise the estimated value for @xmath12 would be substantially _ underestimated_. to avoid the complicated issue of possible non - linear biasing on small scales , one can draw constraints on @xmath12 from a comparison of the measured pairwise velocity dispersion with our predictions for cdm models of various @xmath12 ( see also mo et al .
1996 ) . as long as one relies on the cobe normalization , our cdm predictions are fairly definite ; for realistic cdm @xmath154 with flattening , fortunately , the ( unknown ) force softening even with @xmath157 would not make any difference on the prediction of @xmath158 ( see 4.2 and fig.[fig : v12cdmr ] ) .
the remaining two parameters which would complicate the comparison between the predictions and observations include the normalized amplitude of the three - point correlation functions which is unlikely to be determined accurately from a small sample of galaxies , and a possible velocity bias of galaxies which may result from the dynamical friction but whose quantitative effect is still controversial in literature .
although the present analysis provides a theoretical improvement over the conventional cvt , the usefulness of the cvt to estimate @xmath12 also relies on the quality of the observational data .
in fact , mo et al .
( 1993 ) noted that the observed velocity dispersions vary fairly significantly from sample to sample , and also are sensitive to the modeling of the data .
we will discuss this topic in details elsewhere ( suto & jing 1996 ) .
we thank houjun mo and gerhard brner for useful discussions and comments on the manuscript , and takahiro t. nakamura for providing the fitting formula for the cobe 2yr normalization .
ypj gratefully acknowledges the receipt of an alexander - von - humboldt research fellowship .
ys is grateful to gerhard brner for hospitality at max - planck - institut fr astrophysik , where the present work was done .
this research was supported in part by the grants - in - aid by the ministry of education , science , sports and culture of japan ( 07ce2002 ) to resceu ( research center for the early universe ) .
ballinger , w.e . , peacock , j.a . &
heavens , a.f .
1996;mnras , in press ; bartlett , j. g. & blanchard , a. 1996;a&a;307;1 ; davis , m. & peebles , p.j.e .
1983;apj;267;465 ; fisher , k.b . , davis , m . , strauss , m.a . , yahil , a. , & huchra , j. 1994;mnras;267;927 ; hamilton , a.j.s . , kumar , p. , lu , e. , & matthews , a. 1991 ; apj;374;l1 ; hamilton , a.j.s .
1992;apj;385;l5 ; jain , b. , mo , h.j .
, & white , s.d.m .
1995;mnras;276;l25 ; jing , y.p .
& brner , g. 1996;a&a , in press ; kaiser , n. 1987;mnras;227;1 ; marzke , r.o . ,
geller , m.j .
, da costa , l.n . , & huchra , j .
1995 ; aj;110;477 ; matsubara , t. & suto , y. 1994;apj;420;497 ; matsubara , t. & suto , y. 1996;apj october 10 issue , in press ; mo , h.j . ,
jing , y.p . , & brner , g. 1993;mnras;264;825 ; mo , h.j . ,
jing , y.p . , & brner , g. 1996;mnras , submitted ; nakamura , t.t .
1996 ; master thesis to the university of tokyo , unpublished ; peacock , j.a . & dodds , s.j . 1994;mnras;267;1020 ; peacock , j.a . & dodds , s.j . 1996;mnras;280;l19 ; peebles , p.j.e .
1976;astrophys.sp.sci.;45;3 ; peebles , p.j.e .
1980 ; the large scale structure of the universe;princeton university press;princeton ; sugiyama , n . , 1995;apjs;100;281 ; suto , y. 1993;prog.theor.phys.;90;1173 ; suto , y. 1996;in proceedings of the inauguration conference of asia pacific center for theoretical physics;eds .
j.w.kim;world scientific;singapore;in press ; suto , y. & jing , y.p .
1996;in preparation ; suto , y. & matsubara , t. 1994;apj;420;504 ; suto , y. & matsubara , t. 1996;apj , submitted ;
|
we discuss the effect of the finite size of galaxies on estimating small - scale relative pairwise peculiar velocity dispersions from the cosmic virial theorem ( cvt ) .
specifically we evaluate the effect by incorporating the finite core radius @xmath0 in the two - point correlation function of mass , i.e. @xmath1 and the effective gravitational force softening @xmath2 on small scales .
we analytically obtain the lowest - order correction term for @xmath3 which is in quantitative agreement with the full numerical evaluation . with a nonzero @xmath2 and/or
@xmath0 the cosmic virial theorem is no longer limited to the case of @xmath4 .
we present accurate fitting formulae for the cvt predicted pairwise velocity dispersion for the case of @xmath5 . compared with the idealistic point - mass approximation ( @xmath6 )
, the finite size effect can significantly reduce the small - scale velocity dispersions of galaxies at scales much larger than @xmath2 and @xmath0 . even without considering the finite size of galaxies , nonzero values for @xmath0
are generally expected , for instance , for cold dark matter ( cdm ) models with a scale - invariant primordial spectrum . for these cdm models ,
a reasonable force softening @xmath7 would have rather tiny effect .
we present the cvt predictions for the small - scale pairwise velocity dispersion in the cdm models normalized by the cobe observation .
the implication of our results for confrontation of observations of galaxy pair - wise velocity dispersions and theoretical predictions of the cvt is also discussed . # 1 ( # 1 )
# 1\ { # 1 } # 1 # 1#1 0.1 in the university of tokyo + 0.1 in research center for the early universe + resceu-33/96 + utap-240/96 +
|
the unified model of active galactic nuclei ( agns ) @xcite assumes the existence of dust torus that surrounds the central hierarchical structure consisting of a super massive black hole , an accretion disk , and a broad line region ( blr ) .
the dust grains in the torus absorb the uv / optical continuum emission from the accretion disk and re - radiate in the near infrared ( ir ) wavelength region with some lag time corresponding to the light travel time from the accretion disk to the inner radius of dust torus .
since the heated dust eventually sublimates at a constant temperature of 1500 - 1800k , a more luminous agn should have a larger dust torus and hence a larger lag time , yielding a correlation of @xmath5 between lag time @xmath2 and agn luminosity @xmath3 @xcite .
in fact , based on the long - term multicolor monitoring data from the magnum project @xcite and available archival data , @xcite recently reported such a correlation from a sample of agns spanning a wide range of absolute @xmath4-magnitude from @xmath6 to @xmath7 .
the optical luminosity is a good indicator of uv luminosity , because their variations are well synchronized with each other @xcite .
therefore , the reported correlation in the @xmath4 band indicates that the inner radius of dust torus scales with the uv luminosity as well , which provides a strong piece of evidence for dust reverberation proceeding in the central region of agns .
given the above result of scaling among agns of different luminosities , it is possible to examine how @xmath2 would scale with changing @xmath3 in an individual agn .
this is currently an only diagnosis on its inner structure which is too small to be resolved by direct imaging .
however , since a range of flux variation in an agn is as small as @xmath8 mag , detection of a change of @xmath2 is indeed a challenge and needs an elaborate method that enables precise estimation of @xmath2 .
we here attempt to detect , if any , a change of @xmath2 for the case of a nearby seyfert 1 agn of ngc 4151 , using the six year log accurate and frequent monitoring data for bright agns in the program of magnum . in section [ sec : app ]
, we show details of observation of ngc4151 . in section [ sec : mod ] and section [ sec : tes ] we formulate estimation of @xmath2 and @xmath3 , by making use of model light curves . in section [ sec : ana ] we apply the formulation to the case of ngc 4151 and derive its time - changing track in the @xmath2 versus @xmath3 diagram . in section [ sec : dis ] we discuss time scale of dust formation in the central region of ngc 4151 .
ngc4151 is one of the bright agns most frequently observed since 2001 by the magnum telescope . the monitoring period analyzed in this letter covers a length of @xmath9 d from mjd @xmath10 to @xmath11 ( 2001 - 2006 ) , and photometric images were obtained for about 220 nights in the @xmath4 and @xmath12 bands .
the observations , photometry , and data analysis are mostly in common with those in @xcite , and we do not repeat them except for some important basics below .
each night we took images of the agn and the reference star alternatively using the multicolor imaging photometer ( mip ) with the viewing angle of @xmath13 , and performed relative photometry of the agn with an aperture size of @xmath14 .
this size was chosen to minimize the photometric error and the flux variation in the aperture by the seeing effect .
the flux in the aperture contains the flux from the host galaxy , especially from its bulge .
the host component was estimated by the model fitting to its surface brightness as @xmath15 @xcite and @xmath16 @xcite , and we subtracted these contributions from the @xmath4 and @xmath12 light curves , respectively .
the narrow line flux was also significant in the @xmath4 band for ngc4151 , so we subtracted @xmath17 mjy following @xcite .
furthermore , the ir flux contains the flux from the accretion disk , which makes @xmath2 systematically shorter than the actual lag . according to @xcite ,
the ir disk component should vary to synchronize with the optical variation that is originated genuinely from the accretion disk .
this component in the @xmath12 band was then estimated as @xmath18 , and we also subtracted this contribution from the @xmath12 light curve by assuming @xmath19 for the disk emission between the @xmath4 and @xmath12 bands .
the resulting @xmath4 and @xmath12 light curves of ngc 4151 from 2001 to 2006 are shown in figure [ fig : fig2 ] .
the @xmath4 light curve shown here is consistent with the optical continuum light curves reported by @xcite .
we see that the optical flux of ngc 4151 reached a minimum around mjd @xmath10 and then increased until reaching a maximum at mjd @xmath20 .
thereafter , it turned to decrease again to a minimum at mjd @xmath21 .
apparently , the flux variation in the @xmath12 band followed that in the @xmath4 band with clear time delay .
the light curve shown in figure [ fig : fig2 ] is complex , with many peaks and valleys . from it
we need to extract a table of @xmath2 as a function of continuum flux which has sufficient precision for us to test the hypothesis that @xmath2 is proportional to @xmath22 as the flux varies , and/or to understand any difference from that expected result .
the techniques used previously are not adequate . in our series of papers on dust reverberation in agns @xcite
, we estimated @xmath2 and its error based on the cross correlation function ( ccf ) analysis by making use of the structure function ( sf , * ? ? ? * ) for the interpolation of optical flux variability during the monitoring period . in the equal sampling ( es )
scheme of interpolation which we used for the ccf analysis @xcite , the number of data points generated at equal intervals of time can exceed that actually observed . in such a case ,
the simulated light curve shows larger flux variation compared to an overall trend seen from the observed light curve , especially when the data points are generated far from the location of observed epochs .
this necessarily gives rise to too large an error of @xmath2 .
we here adopt the bi - directional interpolation ( bi ) scheme , where all @xmath4-band flux data are paired with @xmath12-band flux data and at least either @xmath4 or @xmath12 flux in each pair is measured by actual observation .
while the bi scheme overcomes a problem of generating too many data points for ccf calculations , the ccf still shows an unreal peak with increasing the number of data pairs .
this drawback is remedied by estimating @xmath2 from the ccf centroid around the peak which is interpreted as a luminosity - weighted distance of the dust torus from the agn central energy source @xcite . as a practice of determining the ccf range for centroid estimation ,
the threshold is usually set to be 0.8 times the ccf - peak value .
if the threshold is larger than @xmath23 , @xmath2 is influenced by an unreal peak of the ccf .
on the other hand , if it is smaller than @xmath24 , @xmath2 is necessarily influenced by sub - features such as a flared foot of the ccf .
we so far used a simple mean of observed @xmath4-band fluxes as an indicator of optical luminosity that corresponds to @xmath2 , but this could seriously be biased by the two factors .
first , the flux mean depends on the sampling frequency of light curve during the monitoring period .
when an agn is often observed in the bright phase rather than in the faint phase , the simple flux mean preferentially gives a value of brighter flux .
second , the flux mean depends on the shape of light curve .
when an agn undergoes an abrupt burst on a flat light curve , the simple flux mean gives a value which is more or less the same as the value of flat part of the light curve , although the flat part contributes very little to the value of cross - correlation ( cc ) coefficient for the estimation of @xmath2 .
considering above , we take a mean of @xmath4-band fluxes generated at equal intervals of time , after excluding the data which statistically show no flux variation in time bins . to pick up such data
, we use the equally interpolated optical light curve for which the size of time bins is taken as the mean interval of monitoring observations . the variability in each bin
is tested by @xmath25 with a significance level of @xmath26 , and the data which are judged as showing no flux variation are excluded in calculating the mean of @xmath4-band fluxes .
the distribution of this flux mean for all simulated light curves is then used to derive an average optical flux @xmath27 and its error .
we next use model light curves to verify that our analysis procedure reliably finds a characteristic lag @xmath2 and its associated flux @xmath28 for each of the individual variability events that combine to make the complex light curve shown in figure 1 . in order to quantify the above estimation of a characteristic lag @xmath2 and its associated flux @xmath27 as well as their errors , we model the light curves to calculate ccfs . given a @xmath4 light curve @xmath29
, we construct a corresponding @xmath12 light curve as @xmath30 , where @xmath31 for @xmath32 which roughly reproduce the lags seen in figure [ fig : fig2 ] . here , the units are days in time and arbitrary in flux . then we generate 30 data points along each of the @xmath4 and @xmath12 light curves , and disperse the points according to a gaussian distribution of photometric errors taken from the actual observations by the magnum telescope .
the models include three characteristic shapes of ( 1 ) linear bottom , ( 2 ) quadratic bottom , ( 3 ) bursting gaussian peak , as shown in order from top to bottom in the left column of figure [ fig : fig1 ] .
the models 1 to 3 are shown in figure [ fig : fig1 ] to examine the dependence on shape of time - symmetric incident @xmath4 light curve used for ccf calculation .
in addition to all these models with a standard index of @xmath33 , we consider two cases of @xmath34 and @xmath35 for each of light curve shapes 1 and 2 .
we note , however , that @xmath36 is considered only for academic purpose and contradicts with the geometrical @xmath37-dilution of the incident flux from the central energy source .
following @xcite , we applied the bi method to each of model light curves to calculate the ccf and also the cc centroid distribution ( histogram ) by monte - carlo simulations , as shown in the middle column of figure [ fig : fig1 ] for @xmath38 .
the results of @xmath2 and its error are given on the panels of this column and in table [ tab : tab1 ] .
the results of @xmath2 for the models 1 to 3 give values around @xmath39 d. this indicates that @xmath2 could be estimated independently of the shape of light curve .
now we derive an average flux @xmath27 for each of the models 1 to 3 , following the way described in section [ sec : mod ] .
the horizontal arrows in the left column of figure [ fig : fig1 ] specify the periods in which the data are rejected in the estimation of flux mean due to their invariability .
we calculate the distribution of this flux mean for the light curves generated by monte - carlo simulations , and present it as a histogram from which @xmath27 and its error are derived , as shown in the right column of figure [ fig : fig1 ] for @xmath38 .
the results of @xmath27 and its error are given on the panels of this column and in table [ tab : tab1 ] . to be consistent with the measured value of @xmath2
, we compare our estimate of @xmath27 with that from the relation @xmath31 adopted in the ccf analysis .
the relative difference of @xmath40 is shown in table [ tab : tab1 ] .
we see that @xmath41 for @xmath33 is less than @xmath42 for all the models 1 to 3 , which is considerably improved over our previous estimate of @xmath43 by using the simple flux mean .
the result of @xmath27 for different values of @xmath34 and @xmath35 tabulated in table [ tab : tab1 ] shows little difference from that for @xmath33 , which indicates that estimated @xmath27 is almost independent of this parameter .
in addition , various other light curves in asymmetric shapes also give an essentially same result .
given that the consistent estimation of @xmath2 and @xmath27 has been confirmed as above by using the model light curves , we apply it to the real data for ngc 4151 .
we divided the whole monitoring period into eight separate periods , each of which contains a single feature of flux variation , as above , with a sufficient number of more than 40 flux data pairs for the ccf analysis .
the middle date of each period is taken as an epoch that represents each period . following the procedure described in section [ sec : mod ]
, the monte - carlo simulation was repeated 500 times for each period to form the distribution of cc centroid lag from which @xmath2 was obtained as the median and its error as @xmath44 34.1 percentile from the median .
similarly , we calculated the distribution of flux mean from which @xmath27 and its error were obtained .
their results for eight separate periods are given in table [ tab : tab2 ] as well as in figure [ fig : fig3 ] .
we here note that the @xmath12-band data of ngc 4151 were well sampled and the observed @xmath12 light curves were smooth enough .
therefore , in order to avoid additional ir flux variation which consequently overestimates the errors in @xmath2 , use of the sf for the interpolation was turned off and the simple linear interpolation was adopted for the @xmath12-band data in making pairs with the @xmath4-band data .
it is apparent that @xmath2 is not constant in time . for the most extreme case
, the difference of @xmath2 between the earlier periods of 1 through 4 and the later periods of 6 through 8 is larger than the measurement error by many @xmath45s .
since @xmath2 corresponds to a light - travel distance to the surrounding dust from the agn center , this should be the first convincing evidence that the inner radius of dust torus did change in an individual agn ( cf .
* ; * ? ? ?
the change of inner radius of dust torus should occur reflecting the variation of the incident uv / optical flux from the central energy source .
some theoretical models suggest the existence of sublimation radius inside which dust grains are sublimated @xcite .
the sublimation radius expands when the uv / optical continuum becomes bright , then it retreats when the continuum becomes faint .
if dust grains were sublimated or replenished immediately after the uv / optical flux variation , like the model we assumed in section [ sec : mod ] , the expected changes of @xmath2 and @xmath27 should trace the simple relation of @xmath5 which has the slope indicated by the thick line in figure [ fig : fig3 ] .
we see from this figure that the observed change of @xmath2 follows the variation of @xmath27 in overall trend as expected .
these results indicate that the dust torus is not a distinct , separate physical structure but is a part of continuous structure starting from the blr component out to dust component with its inner dust edge set by the sublimation radius that changes with the uv / optical flux variation @xcite .
on the other hand , however , the observed time - changing track does not exactly trace the simple relation of @xmath5 .
the lag @xmath2 at the period 1 is well above that expected for @xmath46 , and does not change so much with brightening of about 1 mag from the period 1 to 3 .
then , with subsequent fading from the period 3 to 6 , @xmath2 still remains constant until the period 4 , and starts to rapidly decrease thereafter .
these deviant behaviors suggest that the inner dust torus did not reach an equilibrium state immediately after the uv / optical flux variation . in the case of period 1 ,
the light curves in figure [ fig : fig2 ] indicate that there was a brighter period before the period 1 .
the expanded inner radius of dust torus at the brighter period could still remain larger than the radius that would be expected at the period 1 . in this context , the behavior of almost constant @xmath2 followed by its rapid decrease in the fading period places a constraint such that a time scale of dust replenishment in the central region of agn should be as long as the elapsed time of about one year from the period 4 through 6 .
a promising mechanism of such replenishment is re - formation of dust grains rather than their infall from the outer region .
this is understood by considering that the decrease of @xmath2 from the period 4 to 6 is 24.5 d or a light - travel distance of @xmath47 km in the interval of 309 d between the middle dates of these periods .
if such decrease of @xmath2 would occur with redistribution of dust grains supplied from the outer region of dust torus , the infall velocity would be @xmath48 km s@xmath49 or 7.9% of the light velocity .
this seems highly unlikely , when we compare this infall velocity with a few @xmath50 km s@xmath49 for the velocity dispersion of blr clouds which exist just near the inner dust torus .
consequently , we conclude that re - formation of dust grains did occur in the central region where they had been sublimated . after the dust re - formation period , @xmath2 turned almost constant again from the period 6 through 8 .
this indicates that dust sublimation radius became larger than actual inner radius of dust torus and dust sublimation would start .
the length of period of constant @xmath2 suggests that a time scale of dust destruction could be longer than about one year .
exact measurement of this time scale might have been possible if the light curves were obtained well after the period 8 .
the formulation for improving the accuracy of lag and luminosity measurements was applied to the light curves for eight separate periods containing single features .
this formulation , when applied to the light curves for the whole monitoring period containing several features , gives an almost average of their respective @xmath2 estimates derived from single features , and is still useful in discussing an overall lag
luminosity relationship for a sample of many agns with a wide range of luminosity .
we thank the staff at the haleakala observatories for their help with facility maintenance .
this research has been supported partly by the grants - in - aid of scientific research ( 10041110 , 10304014 , 11740120 , 12640233 , 14047206 , 14253001 , 14540223 , and 16740106 ) and the coe research ( 07ce2002 ) of the ministry of education , science , culture and sports of japan .
antonucci , r. , 1993 , , 31,473 barvainis , r. , 1987 , , 320 , 537 barvainis , r. , 1992 , , 400 , 502 edelson , r. a. et al .
, 1996 , , 470 , 364 koratokar , a. p. , & gaskell , c. m. , 1991 , , 75 , 719 minezaki , t. , yoshii , y. , kobayashi , y. , enya , keigo . ,
suganuma , m. , tomita , h. , aoki , t. , & peterson , b. a. , 2004 , , 600 , l35 nenkova , m. , sirocky , m. m. , nikutta , r. , ivesi , . , & elitzur , m. , 2008 , , 685 , 160 oknyanskij et al .
1999 , astronomy letters , 25 , 483 shapovalova , a. i. et al .
2008 , , 486 , 99 suganuma , m. et al . , 2006 , , 639 , 46 tomita , h. et al . 2006 , , 652 , l13 urry , c. & padovani , p. 1995
, , 107 , 803 white , r. j. , & peterson , b. m. , 1994 , , 106 , 879 yoshii , y. , 2002 , in new trends in theoretical and observational cosmology , ed . k. sato & t. shiromizu ( tokyo : universal academy press ) , 235 yoshii , y. , kobayashi , y. , & minezaki , t. , 2003 , , 202 , 38.03 and @xmath12 light curves of ngc 4151 ( dots ) in units of mjy along the vertical axis on the left , and the lag times @xmath2 ( crosses ) in units of days along the vertical axis on the right . the horizontal bar to each cross
does not specify the error but the time interval for which @xmath2 is estimated , and the vertical bar indicates the error of @xmath2 by the ccf simulations .
the number beside each cross refers to the serial number given to each monitoring period in table [ tab : tab2 ] . ] and @xmath12 light curves ( left column ) , and the corresponding distributions of cc centroid lag ( middle column ) and mean @xmath4 flux ( right column ) , for the cases of time - symmetric incident @xmath4 light curve .
three panels from left to right in each row shows a procedure of estimating @xmath2 and @xmath27 from each model of light curve ( see text ) .
the solid lines on the panels of middle column show the ccfs with their values on the right axis .
the horizontal arrows on the panels of left column indicate the periods for which the flux data are excluded for estimation of @xmath27 .
note that the units are arbitrary in flux . ]
-magnitude diagram , as shown by the dot - connecting thin line .
the number beside each dot refers to the serial number given to each monitoring period in table 2 .
the thick line is the scaling relation of @xmath5 expected from dust reverberation .
we used @xmath51 km s@xmath49 mpc@xmath49 for estimation of @xmath46 . ] cllllr 1 & linear bottom & 0.5 & @xmath52 & @xmath53 & @xmath54 + 2 & quadratic bottom & 0.5 & @xmath55 & @xmath56 & @xmath57 + 3 & gaussian peak & 0.5 & @xmath58 & @xmath59 & @xmath60 + 1a & linear bottom & 0.2 & @xmath61 & @xmath62 & @xmath63 + 2a & quadratic bottom & 0.2 & @xmath64 & @xmath65 & @xmath66 + 1b & linear bottom & 0.8 & @xmath67 & @xmath68 & @xmath57 + 2b & quadratic bottom & 0.8 & @xmath69 & @xmath70 & @xmath71 [ tab : tab1 ] clll 1 & @xmath72@xmath73 & @xmath74 & @xmath75 + 2 & @xmath76@xmath77 & @xmath78 & @xmath79 + 3 & @xmath80@xmath81 & @xmath82 & @xmath83 + 4 & @xmath84@xmath85 & @xmath86 & @xmath87 + 5 & @xmath88@xmath89 & @xmath90 & @xmath91 + 6 & @xmath92@xmath93 & @xmath94 & @xmath95 + 7 & @xmath96@xmath97 & @xmath98 & @xmath99 + 8 & @xmath100@xmath101 & @xmath102 & @xmath103 [ tab : tab2 ]
|
the long - term optical and near infrared monitoring observations for a type 1 active galactic nucleus ngc 4151 were carried out for six years from 2001 to 2006 by using the magnum telescope , and delayed response of flux variations in the @xmath0 band to those in the @xmath1 band was clearly detected .
based on cross correlation analysis , we precisely measured a lag time @xmath2 for eight separate periods , and we found that @xmath2 is not constant changing between 30 and 70 d during the monitoring period . since @xmath2 is the light travel time from the central energy source out to the surrounding dust torus , this is the first convincing evidence that the inner radius of dust torus did change in an individual agn . in order to relate such a change of @xmath2 with a change of agn luminosity @xmath3 , we presented a method of taking an average of the observed @xmath4-band fluxes that corresponds to the measured value of @xmath2 , and we found that the time - changing track of ngc 4151 in the @xmath2 versus @xmath3 diagram during the monitoring period deviates from the relation of @xmath5 expected from dust reverberation .
this result , combined with the elapsed time from period to period for which @xmath2 was measured , indicates that the timescale of dust formation is about one year , which should be taken into account as a new constraint in future studies of dust evolution in agns .
|
we first consider an effective continuum model of the surface states of a ti and their coupling to local magnetic impurities using the hamiltonian @xmath6 here @xmath7 is the fermi velocity of the surface states , @xmath8 denotes the vector of pauli matrices , and @xmath9 is the electron annihilation spinor at momentum @xmath10 .
the magnetic impurities are modeled by the total scattering potential @xmath11 , which includes both scattering off a potential @xmath3 , with the identity matrix @xmath12 , and magnetic moment @xmath13 , both acting as point defects at positions @xmath14 . for simplicity ,
since the quantum nature of the spins is not crucial , we use large spin moments @xmath15 and weak couplings @xmath16 , requiring @xmath17 constant , such that the impurity spins can be treated as classical . with the spin of the surface states oriented within the @xmath18-plane in @xmath19 , a magnetic field along
the @xmath20-direction gaps the surface spectrum at the dirac point , as it adds a term proportional to @xmath21 @xcite .
single magnetic impurities with moment along @xmath20 ( with potential scattering ignored ) and with a finite spatial extent have also been shown to give an effective local gap @xcite , while a single point - like magnetic impurity results in no incipient gap @xcite . for finite concentration of point - like magnetic impurities
, their spins has been shown to align and thus produce an effective magnetic field @xcite , and we are primarily interested in this latter system . to proceed , we study the scattering off a finite concentration of impurities in eq . using the @xmath22-matrix approach , see e.g. refs .
@xcite . for a low density of localized magnetic impurities ,
all with the scattering potential @xmath23 , the impurity averaged fermion green function ( gf ) reads @xmath24^{-1}$ ] to first order in the density of impurities @xmath25 . here
, the self - energy in the born approximation is given by @xmath26^{-1}\bfv_0 $ ] , with @xmath27 .
resolving the algebra , we obtain [ eq - gba ] @xmath28\sigma_0+[\hbar v_f \bfk\times\hat{\bf z}+\widetilde\bfm(z)]\cdot\bfsigma } { [ z-\widetilde{u}(z)]^2-\widetilde m_z(z)-|\hbar v_fke^{i\varphi}+\widetilde m_+(z)|^2 } , \label{eq - gk } \\
\widetilde{u}(z)= & \rho\frac{u - g_0(z)[u^2-m^2]}{[1-g_0(z)u]^2-g_0 ^ 2(z)m^2 } , \label{eq - u } \\
\widetilde{\bfm}(z)= & \frac{\rho}{[1-g_0(z)u]^2-g_0 ^ 2(z)m^2 } \bfm .
\label{eq - delta}\end{aligned}\ ] ] here @xmath29\sigma_0/4\pi(\hbar v_f)^2 $ ] , where we have introduced a finite cut - off energy @xmath30 for the band width of the surface states .
in addition , we have introduced @xmath31 , @xmath32 , @xmath33 , and @xmath34 . . here , @xmath35 , @xmath36 , @xmath37 , and @xmath38 .
] using @xmath39 to calculate the low - energy band structure , the formulation provided in eqs .
( [ eq - gba ] ) enables a continuous variation of the potential scattering @xmath3 . in fig .
[ fig - nk ] we plot the evolution of the calculated band structure for increasing potential scattering @xmath3 , but fixed strength of the local magnetic impurity moment @xmath40 .
we here explicitly choose a magnetic moment such that there is a clear gap at the dirac point for zero potential scattering .
the band structure shows an essentially unperturbed and gapped band structure for potentials @xmath41 , see figs .
[ fig - nk](a ) and ( b ) .
however , the gap vanishes rapidly for increasing @xmath42 , as seen in figs .
[ fig - nk](b ) ( h ) .
the flattening of the lower part of the dirac cone for @xmath42 indicates the presence of an impurity resonance .
even though the impurity resonance is located well below the dirac point , there is enough spectral weight available from its tail to significantly modify the gap and for @xmath43 the gap is completely filled .
we thus conclude that even weak potential scattering significantly alters the low - energy spectrum by filling the magnetically induced gap at the dirac point .
there is also an overall upward energy shift of the band structure with increasing @xmath3 , indicated by the shift @xmath44 in the denominator of @xmath45 in eqs .
this is expected since potential scattering contributes as a local positive ( hole ) doping .
this effect is advantageous since it both breaks particle - hole symmetry , not usually present in tis @xcite , and it can also be seen to mimic additional carrier doping of the material .
both the gap turning into a region of suppressed , but finite , intensity and the energy shift in fig .
[ fig - nk ] agree very well with the experimental results reported in ref .
the results in fig .
[ fig - nk ] are for magnetic moments along @xmath21 . adding a small moment also along @xmath46 does not change the results .
this is clear since the shift @xmath47 in the dispersion relation in eqs .
[ eq - gba ] does not act as a mass term in the hamiltonian , but merely renormalizes the angular dependence of the energy . in order to compare to local probing experiments and
more clearly resolve the impurity resonance structure , we are also interested in the integrated dos around the fermi energy @xmath48 . for this purpose
, we study the properties of @xmath49 . assuming @xmath50 , there is only a weak angular dependence in the momentum summation and we can perform the summation analytically , giving @xmath51\sigma_0\!+\!\widetilde\bfm(z)\
! \cdot \ !
\bfsigma}{4\pi(\hbar v_f)^2 } \log\frac{|\widetilde\bfm(z)|^2-(z-\widetilde{u}(z))^2}{d_c^2 } .\end{aligned}\ ] ] the resulting dos is plotted in fig .
[ fig - n ] for different ratios @xmath52 . in the case of vanishing @xmath3 ,
[ fig - n](a ) , the dos retains the linear dispersion of the surface states and has a distinct energy gap centered around the dirac point , as is expected for a finite concentration of purely magnetic impurities @xcite . slightly increasing @xmath53
, there is an overall energy shift of the spectrum towards a hole doped state , but more importantly , the size of the energy gap also decreases .
the reduced gap is a direct consequence of the impurity resonance from the potential scattering moving toward lower energies and effectively crowding out the gap .
already for @xmath54 , a very realistic value for the potential scattering of a magnetic impurity atom ( see concluding remarks ) , there is a clear resonance peak visible in the low - energy spectrum and its tail has already started to fill up the energy gap .
thus , we again see clearly how including a realistic potential scattering term lifts the magnetically induced gap and , as a consequence , the dos instead tends towards that of a potential impurity @xcite .
this offers a simple explanation to recent experimental studies , which have found no significant difference between magnetic and non - magnetic surface impurities @xcite .
we also note how the impurity resonance at intermediate @xmath3 values tends to split up in two peaks , a feature which has also been reported experimentally @xcite .
( a ) and @xmath55 ( b ) .
other parameters are the same as in fig .
[ fig - nk ] . ]
the continuum model results demonstrate the importance of including the potential scattering contribution for magnetic impurities , since the resulting impurity resonance easily fills the magnetically induced gap .
in fact , a realistic impurity can easily provide a potential scattering perturbation ( @xmath56 ev ) exceeding that of the bulk energy gap in a ti ( @xmath57 ev ) .
this also raises the question if a continuum surface model , with an infinitely large bulk gap , accurately captures the low - energy properties of ti surface impurities .
we therefore also study magnetic surface impurities with finite potential scattering in a full 3d lattice model .
this lattice calculation excellently complements the continuum model , as it not only includes a finite bulk gap , but also captures the nonzero penetration of the surface states into the bulk of the ti .
moreover , we only access the behavior of dirac delta impurities in the continuum model , whereas the lattice model naturally allows for non - singular impurities .
more specifically , we use a simple tight - binding model of a ti , which consists of @xmath58-orbitals arranged on a diamond lattice @xcite : @xmath59 here @xmath60 is the annihilation operator on site @xmath61 in the lattice with spin - index @xmath62 , @xmath63 is the nearest neighbor hopping , @xmath64 is the spin - orbit coupling , @xmath65 is the cubic cell size , and @xmath66 are the two bond vectors connecting next - nearest neighbor sites @xmath61 and @xmath67 .
this system becomes a strong ti with a single dirac surface cone when setting @xmath68 for one of the nearest neighbor directions not parallel to ( 111 ) @xcite . to access a surface ,
we construct a slab in the ( 111 ) direction with abbcc ... aabbc stacking termination . to avoid a hybridization gap between the two slab surfaces , we use eight lateral unit cells , each consisting of six atomic layers . finally , we set @xmath69 and @xmath70 , which gives @xmath71 for the surface states , the same as in @xmath19 .
( a ) and @xmath72 ( b ) . here
@xmath73 and @xmath74 . ] to study surface impurities , we create a rectangularly shaped surface supercell with @xmath75 sites in each direction , resulting in a surface area @xmath76 , where @xmath77 is the unit of length .
we then add @xmath78 to the hamiltonian , where @xmath79 is the total impurity scattering potential and @xmath80 is one specific surface lattice site within the supercell .
we only consider single - site impurities , which model substitutional or on - top absorbed adatoms , as those most straightforwardly connects to the continuum model results .
however , we do not expect any qualitative changes for extended defects .
for the impurity , we include both potential and magnetic scattering : @xmath81 .
the surface states in @xmath82 do not have as simple a spin structure as the continuum model @xmath19 , but we nonetheless find that a zeeman magnetic field along the @xmath83-direction develops a clear energy gap at the dirac point .
we therefore use a total scattering potential @xmath84 to study the influence of potential scattering on the magnetic impurity induced gap .
we solve @xmath85 in the supercell using exact diagonalization and use the eigenstates to calculate the local dos at each lattice site .
we use a @xmath86 @xmath87-point grid to achieve a high energy resolution capable of accurately resolving small energy gaps and a 0.01 gaussian broadening to compensate for the finite system size . in fig .
[ fig : urun ] , we plot the local surface dos maximally away from a @xmath74 impurity in a @xmath73 supercell , corresponding to @xmath88 surface impurity concentration . at @xmath89 , we see a clear energy gap at the dirac point centered around the fermi level . slowly increasing @xmath3 moves the energy gap to slightly higher energies and , at the same time , very clear impurity resonance peaks start to appear at progressively lower energies .
the latter causes the energy gap to shrink and it is completely filled already at @xmath90 . we especially note that the gap disappears already for small values of @xmath3 .
this is very far from the unitary scattering limit , where the impurity resonance peak is firmly centered around the dirac point @xcite and then generates such a large amount of dos around the fermi level , that the system becomes prone to spontaneous magnetization through a stoner - like impurity mechanism @xcite .
instead , the energy gap in fig .
[ fig : urun ] disappears due to the tail of the impurity resonance peak adding spectral weight at the dirac point .
we also note that the impurity resonance has a clear double - peak structure , which we here attribute to bonding and anti - bonding impurity bands . while we in fig .
[ fig : urun ] plot the local dos far away from the impurity , the energy gap is a global property and does not change throughout the surface .
the impurity resonances are naturally taller closer to the impurity , but we find that the peaks are non - dispersive , and thus their influence on the energy gap is the same both near and far away from the impurity .
quite remarkably , all the results and trends with increasing strength of the potential scattering @xmath3 are very similar for the 3d lattice model in fig .
[ fig : urun ] and the continuum surface model in fig .
[ fig - nk ] .
these include the energy gap shrinking and moving to higher energies , before finally disappearing for @xmath52 of the order of one , as well as the two - peaked impurity resonance structure .
this is especially noteworthy since the lattice calculation has a bulk gap of only about @xmath91 and the impurity resonance state has been found to penetrate as deep as ten layers into the ti @xcite . in order to explicitly track the influence of the potential scattering on the energy gap , we plot in fig .
[ fig : extgap](a ) the extracted energy gap as function of @xmath3 for several different impurity concentrations and magnetic moments . for supercell sizes @xmath75 and impurity magnetic moments @xmath92 . dashed line is for chemical potential @xmath93 . ] clearly , the energy gap is larger for larger @xmath92 and higher impurity concentrations , as both generate an overall larger effective zeeman magnetic field .
however , we find that the gap in all cases diminishes and is finally completely filled for very realistic strengths of the potential scattering associated with a magnetic impurity .
this result is unaffected if we add a finite chemical potential @xmath94 to the hamiltonian @xmath82 ( dashed line ) .
thus both intrinsic doping and local charge puddles @xcite will not affect the results , if anything , non - local effects from charge puddles have been proposed to further diminish the gap @xcite . in fig .
[ fig : extgap](b ) , we finally plot the position of the dirac point . there is a sub - linear increase in position of the dirac point with increasing @xmath3 , but , notably , the size of the magnetic moment does not influence the position of the dirac point . adding a finite @xmath94 simply shifts the dirac point an equivalent distance .
finite concentration of magnetic impurities , where a zeeman magnetic field is produced as a collective effect @xcite , has been considered to be a promising pathway for gap opening and thus functionalizing the ti surface .
we have here shown that by also including the potential scattering , present for all impurities , the magnetically induced gap can be completely filled .
potential scattering in dirac materials is known to give rise to impurity resonances @xcite , and we find that the tail of their spectral weight easily fills up the energy gap .
most strikingly , we find that the energy gap is completely filled for a ratio between the potential and magnetic scattering contributions as low as 1 10 , which is well within the range of expected values for the potential ( 1 10 ev ) and magnetic ( 0.1 1 ev ) scattering for magnetic dopants in tis .
the remarkably close alignment between results from a continuum surface model and a 3d tight - binding lattice model makes it possible to rule out modeling deficiencies .
our results therefore provide new crucial understanding and could offer to resolve the seemingly contradictory experimental situation as to if magnetic impurities induce an energy gap in the ti surface states .
|
we show that the energy gap induced by ferromagnetically aligned magnetic impurities on the surface of a topological insulator can be filled , due to scattering off the non - magnetic potential of the impurities . in both a continuum surface model and a three - dimensional tight - binding lattice model
, we find that the energy gap disappears already at weak potential scattering as impurity resonances add spectral weight at the dirac point .
this can help explain seemingly contradictory experimental results as to the existence of a gap .
the recent discovery of topological insulators ( tis ) @xcite has led to exciting possibilities for producing electronically engineered states .
tis are bulk insulators but host metallic surface states with a dirac - like dispersion @xcite .
the momentum - spin locking in the surface states @xcite offers unique technological capabilities , especially if an energy gap can be created .
a gap in the ti surface states can be generated by time - reversal breaking perturbations and one of the currently most discussed approaches for engineering a gap is magnetic doping @xcite .
the idea is that ferromagnetically ordered impurities will produce a net magnetic field , which then gaps the ti surface states @xcite . beyond an energy gap
, this would also give rise to multiple other exotic phenomena , such as the anomalous hall effect recently observed @xcite .
however , despite a multitude of recent experimental studies , evidence for a gap in the ti surface states from magnetic doping remains controversial .
both angle - resolved photoemission spectroscopy ( arpes ) and scanning tunneling spectroscopy ( sts ) measurements have reported the presence of an energy gap @xcite , while other similar studies have found no gap @xcite .
interestingly , several studies have even reported no significant difference between magnetic and non - magnetic surface impurities @xcite .
the key assumption , when expecting an energy gap from magnetic impurities , is that the ti surface electrons only see an average zeeman magnetic field .
here we point out that magnetic impurities , such as fe or cr , also strongly scatter electrons .
thus , in addition to an effective magnetic field , the presence of magnetic impurities also adds non - magnetic , or potential , scattering .
potential scattering is known to induce low - energy impurity resonances in dirac materials @xcite , ranging from graphene @xcite and @xmath0-wave superconductors @xcite , to tis @xcite . in this work ,
we investigate the effect on the ti surface states of both the magnetic and potential scattering by magnetic impurities .
we find that there are two effects that simultaneously modify the dirac spectrum of the ti surface states .
first , the presence of magnetic scattering opens a gap in the spectrum .
the gap is generated due to the magnetic scattering @xmath1 modifying the energy dispersion relation @xmath2 .
secondly , the potential scattering ( @xmath3 ) induces impurity resonance states , which add low - energy states to the spectrum .
the relative strength of the potential and magnetic scattering determines the net density of states ( dos ) . in the case of weak potential scattering ( @xmath4 ) , we find a well - defined gap . on the other hand , for stronger potential scattering ( @xmath5 ) , the tail of the impurity resonances dominates the low - energy spectrum around the dirac point and the gap is filled .
this provides a unifying framework that allows us to reconcile the conflicting claims about the presence / absence of a gap at the dirac point in magnetically doped tis .
more specifically , we use both a continuum surface model and a tight - binding three - dimensional ( 3d ) lattice model of a ti , where we introduce dilute concentrations of magnetic impurities .
we find that these two distinct models generate remarkably similar results and thus produce a convincing picture regarding the role of magnetic and potential contributions in magnetic impurity scattering .
|
topological sigma models , first put forward by witten @xcite , have long fascinated a number of theoretical physicists and mathematicians .
most remarkably , the task of summing up worldsheet instantons is nowadays elegantly formulated by the theory of gromov - witten invariants .
explicit computations of them are still being actively pursued .
it goes without saying that among many possible target spaces calabi - yau threefolds have played distinguished roles and are of lasting interest to string theorists . since the initial appreciation of the significance of d - branes there has been the lingering hope that the gromov - witten theory of calabi - yau threefolds might be completely rewritten in the language of bps d - branes .
this contribution is intended for explaining the picture which , to my eye , looks particularly attractive in this regard .
this is based on the general philosophy : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the geometry of @xmath0-@xmath1 branes ( and _ not _ simply @xmath0-branes ) provides an alternative description of gromov - witten invariants of calabi - yau threefolds .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ what we actually imagine is very simple and intuitive : in analogy with the ( generalized ) super kac - moody algebras we regard the string partition function as the inverse of denominator function " and interpret the bound state degeneracies of @xmath0-@xmath1 branes as the super root multiplicities " .
the gromov - witten potentials are then extracted from the string partition function . in particular , the variable @xmath2 measuring @xmath1-charge is related to the genus expansion parameter @xmath3 of gromov - witten theory by @xmath4 .
this sort of idea was formerly presented in @xcite when the calabi - yau threefold is elliptically fibered over a hirzebruch surface . there , an analogy to borcherds products @xcite was pursued .
recall that borcherds products or their inverses arise in some cases as the denominator functions of generalized ( super ) kac - moody algebras .
the first hint of the relevance of borcherds products to gromov - witten theory was given by harvey and moore @xcite .
i will report further progress on the string partition functions of these elliptic calabi - yau threefolds elsewhere @xcite .
so the first purpose of this work is simply to extend the @xmath0-@xmath1 picture of @xcite to general calabi - yau threefolds focusing on those aspects which are believed to be independent of any particular details of the threefolds .
some time ago gopakumar and vafa @xcite proposed an alternative reformulation of gromov - witten theory of calabi - yau threefolds based on the space - time effective theory interpretation of gromov - witten potentials .
this claim has been influential but at the same time very mysterious ( at least to the author ) .
the second objective of this paper is to discuss how this proposal of gopakumar and vafa may actually reconcile with ours .
quite recently , the relation between singular instantons of @xmath5 @xmath6 gauge theory and gromov - witten theory has been discussed @xcite in relation to the topological vertex formalism @xcite .
this gauge theoretic approach ( donaldson - thomas theory ) is probably a dual viewpoint of our @xmath0-@xmath1 picture in the same way point - like instantons of @xmath7 @xmath6 gauge theory describe @xmath1-branes .
so , as far as the ideology is concerned , these works seem to have some overlaps with @xcite and the present work .
nevertheless , there is a marked difference in practice : they discuss the sum side " with a supply of explicit calculations for local toric calabi - yau threefolds using the localization technique whilst we discuss the product side " inspired from the examples of certain elliptic calabi - yau threefolds and the associated borcherds - like products @xcite@xcite
. it will be very interesting ( and necessary ! ) to investigate the existence of
sum = product " formulas connecting both sides for a general calabi - yau threefold .
in fact , the topological vertex formalism of @xcite seems to allow an intuitive understanding in terms of @xmath0-branes and @xmath1-branes .
[ comment ] for comments on this point .
i am grateful to k. yoshioka for the collaboration in @xcite .
i also thank a. okounkov for kindly pointing out my nonsensical statement in the previous version of the manuscript and for explaining to me the marvelous proposal of him and his collaborators . .
for a rational function @xmath8 of one variable @xmath2 we define @xmath9 as follows : @xmath10 is the laurent series of @xmath8 at @xmath11 and @xmath12 is that at @xmath13 .
consider , for instance , @xmath14 for an integer @xmath15 .
if @xmath16 then @xmath17 and @xmath18 coincide in @xmath19 $ ] .
however , @xmath20\,.\ ] ] a calabi - yau threefold @xmath21 is a complex 3-dimensional smooth projective variety with @xmath22 and @xmath23 .
we assume that @xmath21 is polarized by some ample line bundle .
if one ever wishes to connect gromov - witten theory of calabi - yau threefolds to some sort of of bps @xmath24-brane counting , one might think that @xmath0-branes alone are relevant since gromov - witten theory is concerned with curve counting problems .
however , this is too naive and even misleading .
what needs to be emphasized is that in any attempt of this sort we have to incorporate the effects of @xmath1-branes in addition to @xmath0-branes .
take as an example the case of a resolved conifold , the total space of @xmath25 .
since the @xmath26 in the resolved conifold is rigid and the jacobian of @xmath26 is just a point , the moduli space of @xmath0-brane wrapping around the @xmath26 is also a point .
if @xmath0-branes were the only relevant @xmath24-branes , this would mean that the gromov - witten theory of the resolved conifold was trivial .
this is simply absurd given the result of @xcite . the origin for
the necessity of @xmath1-branes may be roughly as follows .
let us suppose that we are trying to answer the problem of counting curves in a calabi - yau threefold @xmath21
. there could be several different approaches according what we mean by _
counting_. in a crude approach by @xmath24-branes
one may make curves charged " by putting line bundles , ( or more generally , rank one torsion - free sheaves ) on them and then claim , to within signs , the euler - poincar ' e characteristics of the moduli spaces of such sheaves ( regarded as torsion sheaves on @xmath21 ) with fixed @xmath24-brane charges as the numbers of curves " .
apparently , in this naive approach we are concerned with @xmath0-branes alone . in gromov - witten theory , on the other hand , one tries to count holomorphic maps " from worldsheet connected curves to target curves in @xmath21 and regard this as a good substitute of directly counting curves in @xmath21 .
however , this very substitution introduces some well - known complications .
one of them is the so - called multi - covering effect .
this is rather innocuous since we know more or less how to handle it .
another complication , which seems to be more difficult , is related to conformal invariance of the holomorphic map equation and is known as bubbling phenomena . in order to have a nice intersection theory one must compactify the moduli space of holomorphic maps . for this
we have to include degenerate contributions of bubble trees ( bubbling of bubbling of ) .
it should be precisely for this reason that we have to modify the naive @xmath0-brane approach by including @xmath1-branes when one attempts to rewrite gromov - witten theory in terms of @xmath24-branes .
an intuitive picture of @xmath1-branes bound to a @xmath0-brane is that of vortices .
so morally speaking , _ bubble trees turn into vortices_.
suppose that we are given a super kac - moody algebra .
let @xmath27 denote the additive semigroup generated by simple roots .
we write @xmath28 iff @xmath29 .
any @xmath28 is either even or odd .
consider @xmath30 so that @xmath28 is a positive root iff @xmath31 .
we have @xmath32 where @xmath33 is the @xmath34-th adams operation sending @xmath35 to @xmath36 .
the right hand side is the inverse of the denominator function in the product form . in a nutshell ,
what we intend to do is to make the analogy of this relation in studying a calabi - yau threefold @xmath21 .
we express any element of @xmath37 as @xmath38 where @xmath39 and @xmath40 .
denote by @xmath41 the additive semigroup of classes of effective 1-cycles ( holomorpic curves " ) on @xmath21 .
in other words , @xmath41 is the intersection of the mori cone with @xmath42 .
we write @xmath43 iff @xmath44 . similarly ,
for @xmath45 we write @xmath46 iff @xmath47 \quad \text{or}\quad \bigl[\beta=0 \text{\ and\ } j>0\bigr]\,.\ ] ] assume that there is a suitable @xmath48-grading so that any @xmath46 is either even or odd .
now let us suppose in analogy with @xmath49 the existence of a formal sum : @xmath50 where @xmath51 is a formal symbol satisfying @xmath52 . in the proposal below ,
@xmath53 is in @xmath54 rather than in @xmath55 .
this is to match up with the normalization of gromov - witten potentials .
we next introduce the formal truncated .
] free energy @xmath56 where @xmath33 is again the @xmath34-th adams operation sending @xmath57 to @xmath58 . as the inverse of the denominator function "
we are led to consider the following formal product @xmath59 we might call @xmath60 as the formal truncated string partition function . however the expected relation
@xmath61 " is rather problematic since , in the formal sum , @xmath62 runs over all integers when @xmath43 so that powers of @xmath63 are not well - defined in general .
thus @xmath63 and @xmath60 in the above should be interpreted at best as motivating expressions . in the following
we will introduce regularized versions @xmath64 and @xmath65 related by @xmath66 .
( these are better behaved but somewhat lose direct analogy with super kac - moody algebras . ) those who are familiar with borcherds products and their relations to surfaces , say , @xmath67 surfaces will recognize that we are trying to cook up a similar story for calabi - yau threefolds here .
a novel and distinct point is the introduction of @xmath68 in addition to @xmath42 .
i intend to further discuss this similarity with borcherds products in @xcite . for the above analogy to be anything useful we have to know @xmath69 from the geometry of @xmath21 .
our basic expectation is that @xmath70 should be , in some way or another , identified with the ( super ) degeneracy of bounded @xmath0-@xmath1 branes in @xmath21 with a fixed @xmath0-@xmath1 charge @xmath38 .
therefore , what needs to be done is the geometrical understanding of @xmath0-@xmath1 bound systems .
we first recall the gentlemen s agreement @xcite that even - dimensional @xmath24-branes are related to coherent sheaves and their @xmath24-brane charges are determined by mukai vectors @xcite .
we define the @xmath24-brane charge @xmath71 of a coherent sheaf @xmath72 on @xmath21 by @xmath73 $ ] where @xmath74 is the mukai vector of @xmath72 .
we express @xmath71 in the form @xmath75 where @xmath76 .
note that if the @xmath24-brane charge is of the form @xmath77 then @xmath78 is actually in @xmath79 .
( there is no witten effect . ) for instance , suppose that @xmath80 is a smooth irreducible curve of genus @xmath81 and @xmath82 is a line bundle of degree @xmath83 , then @xmath84 , \chi(c , l))=(0,0,[c],d+1-g)$ ] .
this is the @xmath24-brane charge of a @xmath0-brane singly wrapping around @xmath85 bound with @xmath83 @xmath1-branes . in this paper
any @xmath0-brane is always singly wrapping by allowing non - reduced curves . if the @xmath24-brane charge ( or equivalently the mukai vector ) is fixed , the hilbert polynomial is also fixed since @xmath21 is assumed to be polarized .
so it makes sense to consider the moduli space @xmath86 of semi - stable coherent sheaves on @xmath21 with a fixed @xmath24-brane charge @xmath87 . for simplicity
let us suppose that @xmath86 consists only of stable sheaves .
one well - known fact about @xmath86 is that its expected dimension @xmath88 vanishes because of the serre duality where @xmath89 and @xmath90 is the trace free part of @xmath91 . by stability , @xmath92 and by @xmath23
, we have @xmath93 @xmath94 .
the zero - dimensional virtual moduli cycle was constructed in @xcite .
its degree or virtual length @xmath95 serves as the number of sheaves " .
the zariski tangent space of @xmath86 at @xmath72 is given by @xmath96 and the obstruction space is @xmath97 .
hence the serre duality tells us that if @xmath86 is smooth , the obstruction sheaf is the cotangent bundle of @xmath86 and @xmath98 .
one can also consider the hilbert scheme @xmath99 with a fixed @xmath24-brane charge @xmath87 .
as usual , we have @xmath100 by considering ideal sheaves . the moduli space of possibly disconnected @xmath0-branes and @xmath1-branes , which we call the total @xmath0-@xmath1 moduli space , is given by @xmath101 where @xmath78 runs over all possible values .
at first sight seems not to deserve its name .
for instance , if an isolated @xmath0-brane is singly wrapping around a smooth curve @xmath102 , there should exist degrees of freedom in how it wraps , namely the jacobian @xmath103 . on the other hand in the hilbert scheme @xmath85
is allowed to carry only @xmath104 .
a possible resolution of this puzzle may be as follows .
first of all if a @xmath0-brane wrapping singly around @xmath85 is bounded to several @xmath1-branes , the moduli space of this bound system is given by a symmetric product of @xmath85 ( as we recall later ) .
so this part of the moduli space is directly related to the hilbert scheme .
the pure @xmath0-brane moduli space @xmath103 is actually related to the symmetric products of @xmath85 through the abel - jacobi maps .
so , once we start to consider both @xmath0-branes and @xmath1-branes simultaneously , we should forget about @xmath103 and only consider the symmetric products of @xmath85 as fundamental in order not to overcount the degrees of freedom . a more explanation is given below about how pure @xmath0-brane moduli spaces are related to the moduli spaces of bounded @xmath0-@xmath1 branes .
( this is not unrelated to the upcoming interpretation of the gopakumar - vafa proposal . )
a relation between @xmath105 and gromov - witten invariants was already hinted in @xcite .
a connection to the gopakumar - vafa invariant was conjectured in @xcite . recently
, a striking connection of @xmath106 to gromov - witten theory has been proposed in @xcite .
we expect that @xmath69 is related to but exactly describing this relation seems to be a very difficult problem at the moment .
only a limited attempt is given below .
let us call @xmath107 _ of simple class _ _ _ _ _ if its arithmetic genus @xmath108 is non - negative and for each integer @xmath109 there exist a suitable @xmath0-@xmath1 bound moduli space @xmath110 and a @xmath0 moduli space @xmath111 whose properties we spell out in the following .
we assume that @xmath112 is of simple class and set @xmath113 .
the @xmath0-@xmath1 bound moduli space @xmath110 describes a @xmath0-brane wrapping around a curve of class @xmath114 bound to @xmath83 @xmath1-branes .
in particular we expect an intimate connection between @xmath110 and @xmath115 .
we assume that @xmath110 is smooth and connected .
as we will see shortly , it seems natural to have @xmath116 we set @xmath117 for convenience .
this is the moduli space of the _ supports _ of @xmath0-branes .
note that @xmath118 by .
there should exist a morphism @xmath119 and @xmath120 moreover , there will be the universal family @xmath121 which is a flat family of curves of class @xmath114 and of arithmetic genus @xmath108 .
in particular we expect @xmath122 where the morphism @xmath123 is the projection . intuitively speaking , @xmath124 is the family of curves around which @xmath0-branes are singly wrapping . in favorable situations
it is tempting to identify @xmath125 with the relative hilbert scheme of points on curves @xmath126}:=\operatorname{hilb}^d_{\mathcal{c}_\beta/ s_\beta}\,.\ ] ] since we assumed that @xmath110 is smooth , our concern is limited to a smooth @xmath127}$ ] .
let us explain why and seem natural .
we regard @xmath128 as the moduli space of the bound system of a @xmath0-brane and @xmath83 @xmath1-branes in the limit where the support curve of the @xmath0-brane is shrinking to a point .
what does the curve look like ?
since we do not expect any @xmath0-brane degree of freedom in the end , the jacobian of the curve must be a point .
so the curve will be a @xmath26 . with the vortex interpretation given in @xcite and to be recalled later , the moduli space of the @xmath0-brane wrapping around @xmath26 and @xmath83 @xmath1-branes sticked to it
should be given by the @xmath83-th symmetric product of @xmath26 or @xmath129 .
therefore if we take into account the location of the shrinking @xmath26 in @xmath21 , we are led to and .
let us turn to the properties of the @xmath0 moduli space .
we assume that @xmath111 is smooth and connected .
there should exist a natural morphism @xmath130 .
intuitively , @xmath111 is the moduli space of @xmath0-branes singly wrapping around fibers of @xmath121 . we expect an intimate connection between @xmath111 and @xmath131 .
we furthermore expect that @xmath111 is independent of @xmath83 and is isomorphic to @xmath132 .
we should have @xmath133 .
hence @xmath134 is the identity map .
we require the existence of a commutative diagram @xmath135\ar[dr]&\\ \mathfrak{n}_\beta\ar[rr]&&s_\beta}\ ] ] for @xmath112 we have by obvious morphisms . for @xmath136
, @xmath137 may be viewed as taking a section .
if @xmath138}$ ] , one may take @xmath139 where @xmath140 is the relative compactified picard scheme and @xmath141 is the relative compactified jacobian
. then the diagram is replaced by @xmath142}_\beta\ar[dl]\ar[dr]&\\ \bar{\mathcal{j}}_\beta\ar[rr]&&s_\beta}\ ] ] the morphism @xmath127}\to \bar{\mathcal{j}}_\beta$ ] is just the abel - jacobi map .
we should have @xmath143 this ends the list of what we demand for @xmath114 to be of simple class .
we now turn to our proposal on the structure of @xmath69 : by taking into account charge conjugation symmetry , we have @xmath144 where @xmath145 and if @xmath114 is of simple class @xmath146 with @xmath147 this looks like the most natural extension of what we conceived in @xcite .
the reason why we choose the particular sign factor is explained below by comparison with gromov - witten theory .
note that even if @xmath110 is non - empty , @xmath148 can vanish . in our analogy with super kac - moody algebras
this is related to the vanishing of @xmath53 when @xmath38 is not a positive root " . at the moment
we do not know how exactly @xmath149 should be described geometrically when @xmath114 is other than of simple class .
to make contact with gromov - witten theory we further postulate for each @xmath107 there exists a rational function @xmath150 with inversion symmetry @xmath151 such that @xmath152 if @xmath114 is of simple class , we introduce @xmath153 by @xmath154 , namely , @xmath155 moreover we have the expansion of the form @xmath156 where @xmath4 and @xmath157 .
one then observes that @xmath158 so far @xmath51 has been a formal symbol , but in gromov - witten theory we should like to set @xmath159 where @xmath160 is the complexified k " ahler form of @xmath21 .
therefore , motivated by , we introduce @xmath161 and interpret this as a power series expansion in @xmath162 @xmath163 . here
@xmath164 are generators of the mori cone . in addition one
may introduce the truncated free energy by @xmath165 and the truncated string partition function @xmath166 . then ,
if @xmath167 for all @xmath168 and @xmath169 , we have an infinite product representation : @xmath170 where @xmath171 by we have @xmath172 .
hence we have @xmath173 because of @xmath174 . to discuss the relation to gromov - witten theory
we want to set @xmath4 with @xmath175 and @xmath176 very small and interpret @xmath3 as the genus expansion parameter . as above
we always assume @xmath177 for @xmath43 in the following . for such expansions in @xmath3 to be possible , the truncated free energy needs a bit of regularization : @xmath178 let us introduce @xmath179 via the expansion @xmath180 and define @xmath181,\ ] ] where @xmath182 is the euler - mascheroni constant .
the gromov - witten potentials @xmath183 @xmath184 of @xmath21 are given by @xmath185 where @xmath186 [ remweyl ] in seeking analogy with super kac - moody algebras @xmath187 should be interpreted as ( the negative of ) the weyl vector " .
this part is quite delicate and important .
in fact by using we find that @xmath188 where @xmath189 if @xmath190 and @xmath191 .
this is an expected form .
let us give some evidence for our choice .
note first that @xmath192 and @xmath193 this enables us to calculate @xmath194 explicitly . for @xmath195
, we have @xmath196 .
this is familiar to us . as for @xmath197 , the constant map contribution
is correctly reproduced : @xmath198 where @xmath199 is the deligne - mumford moduli stack of stable curves of arithmetic genus @xmath81 without marked points and @xmath200 is the @xmath201-th chern class of the hodge bundle over @xmath199 .
the relation was conjectured physically in @xcite from @xmath202-theory interpretation and in @xcite from duality to heterotic string .
a mathematical proof of was given in @xcite .
an argument here from a purely @xmath0-@xmath1 point of view is new .
suppose that @xmath114 is of simple class and that both @xmath203 and @xmath204 are smooth .
assume that _ all _ the fibers of @xmath121 are smooth irreducible curves of genus @xmath108 .
then the vortex interpretation explained below shows that @xmath205 if @xmath206 we see that @xmath207 .
this is consistent with gromov - witten theory .
not infrequently we encounter non - trivial formulas of the form @xmath208 so far we have been discussing the product side .
it is natural to ask what the sum side looks like .
obviously , the sum should be taken over potentially all the @xmath0-@xmath1 states .
so the total @xmath0-@xmath1 moduli space will be relevant .
quite recently @xcite , in an attempt to understand the formalism of the topological vertex for local calabi - yau threefolds , the connection between singular instantons of @xmath5 @xmath6 gauge theory and gromov - witten theory has been studied .
this is expected to be another description of our @xmath0-@xmath1 picture studied on the sum side . in particular , @xcite treats local calabi - yau threefolds and evaluates , by the localization technique , the generating function of @xmath105 for the moduli spaces of ideal sheaves @xmath209 .
this is a very explicit and remarkable calculation .
in fact the topological vertex formalism @xcite fits quite nicely into the picture of @xmath0-@xmath1 branes . for a local toric calabi - yau threefold ,
the relevant sum is reduced by localization to the one over the torus - fixed configurations of @xmath0-@xmath1 branes . in the approach of @xcite
one writes the diagram consisting of edges and trivalent vertices .
each internal edge corresponds to a @xmath210-fixed @xmath26 .
each vertex corresponds to a @xmath211-fixed point and is the north or south poles of the @xmath26 s of the internal edges emanating from it .
it is clear that the torus - fixed configurations of @xmath0-@xmath1 branes are such that @xmath0-branes are localized to the @xmath210-fixed @xmath26 s ( the internal edges ) and @xmath1-branes are localized to the @xmath211-fixed points ( the vertices ) .
the way several @xmath0-branes are localized to the internal edge @xmath26 s must be treated scheme - theoretically and described by assigning a @xmath212 young diagram to each edge .
similarly , the way @xmath1-branes accumulate to the vertex points must also be treated scheme - theoretically and described by assigning a @xmath213 young diagram to each vertex .
however the configurations of @xmath1-branes and those of @xmath0-branes are not independent .
how these are actually linked is specified by the rule of the topological vertex .
it should be mentioned that having a nice sum expression does not a priori guarantee the connection to gromov - witten theory . for such to exist
sum = product " will probably have to hold .
for super ( generalized ) kac - moody algebras , this problem is related to finding the set of positive roots among @xmath27 , which in general is a subtle and difficult task .
obviously , if our `` product '' proposal and the sum " proposal ( gw / dt correspondence ) of @xcite are consistent , @xmath214 and @xmath215 must be related .
this looks plausible since both are concerned with counting @xmath0 and @xmath1 branes in @xmath21 .
however , the precise geometrical characterization of @xmath214 is yet to be found .
in @xcite , gopakumar and vafa asserted that the @xmath0-brane moduli spaces of a calabi - yau threefold @xmath21 have some characteristic properties .
we considered the @xmath0 moduli space @xmath216 when @xmath114 is of simple class and assumed that @xmath216 is smooth . to recall the gopakumar - vafa proposal ,
let us temporarily lift the smoothness condition of @xmath216 . in this slightly more general context
, we cast their claim taking into account the suggestion of @xcite as follows : let @xmath217 denote the intersection cohomologies of @xmath216 .
there exists a representation @xmath218 in such a way that the restricted representations @xmath219 associated respectively with the first , the second and the diagonal @xmath220 subalgebras of @xmath221 have the following interpretations : @xmath222 resp .
@xmath223 corresponds to the lefschetz action in the fiber resp .
base direction " of the cohomologies of @xmath224 while @xmath225 corresponds to the usual lefschetz action .
hence one can introduce @xmath226\,,\ ] ] where @xmath227 and @xmath228 are respectively the images by @xmath222 and @xmath225 of the cartan generator of @xmath220 .
( the eigenvalues of @xmath227 and @xmath228 are twice spins . )
note the obvious symmetry @xmath229 . according to @xcite ,
this claim is true if @xmath224 is a projective morphism between two normal projective varieties . in the rest of the paper
we only consider the cases when @xmath114 s are of simple class .
hence @xmath110 and @xmath216 do exist and they are smooth and connected .
when dealing with @xmath216 we do not need @xmath230 and @xmath231 will suffice .
gopakumar and vafa argued that @xmath232 s determine the gromov - witten potentials of @xmath21 .
in particular , considering certain combinations of irreducible representations of @xmath220 to be fundamental , they expanded @xmath232 as @xmath233 then they considered that the integers @xmath234 are alternative fundamental invariants here . ] . since @xmath228 acts as the multiplication of @xmath235 on @xmath236 , we have @xmath237 .
this claim of gopakumar and vafa may sound somewhat at odds with our earlier statement : _ considering @xmath0-branes alone is insufficient_. actually they mapped the problem into @xmath202-theory and used a physical interpretation of space - time effective theory typically using schwinger - like computations .
this is the reason why @xmath1-branes secretly sneak in when interpreted in terms of type iia theory .
the consistency of our proposal with that of gopakumar and vafa implies @xmath238 since @xmath239 and the fiber direction " is void , we have @xmath240 .
consequently is true for @xmath112 . despite its innocent looking the geometrical implication
of is rather non - trivial since the interpretations are quite different on both sides of the equation .
before discussing the validity of let us pause to study purely algebraic consequences of .
we expand @xmath241 as and @xmath232 as .
if we set @xmath242 then the following proposition gives the explicit relations between @xmath148 and @xmath234 .
note that since @xmath243 by our definition , the relation @xmath244 below implies @xmath245 .
[ invprop ] let @xmath246 be an integer .
suppose that sequences of numbers @xmath247 and @xmath248 are related by @xmath249 one can express @xmath247 in terms of @xmath248 as : @xmath250 conversely , @xmath248 can be expressed in terms of @xmath251 as : @xmath252 if @xmath247 are integers , so are @xmath248 and vice versa
. implies the following taylor expansion at @xmath11 : @xmath253 it suffices to analyze this equation .
( a ) this is a consequence of straightforward binomial expansions of the right hand side of . since the cases @xmath195 and @xmath254 are obvious , we assume @xmath255 .
notice that is equivalent to @xmath256 then , by the lagrange inversion formula , one finds @xmath257\\ & = \operatorname{res}_{y=0}\left[\frac{1+y}{(1-y)^{2(g - i)-1}}\sum_{d=0}^\infty e_d\ , y^{d - i-1}\right ] , \end{split}\ ] ] where @xmath258 represents the derivative with respect to @xmath2 . to evaluate this residue
it is better to proceed case by case : + suppose that @xmath259 . then we actually have to have @xmath260 and @xmath261.\ ] ] it is straightforward to evaluate this residue .
the result is given by .
+ suppose instead that @xmath262 .
one uses the formula@xmath263 ( valid for a positive integer @xmath34 ) to obtain @xmath264.\ ] ] if @xmath265 one easily sees that @xmath266 . if @xmath267 , the evaluation of the residue gives @xmath268 which can be rewritten as .
this is already obvious from what we have seen in this proof .
we briefly recall the vortex picture of @xmath0-@xmath1 branes . for more on this
see @xcite .
let @xmath102 be a smooth irreducible curve of genus @xmath81 which is rigid in @xmath21 .
suppose that a @xmath0-brane is singly wrapping around @xmath85 and @xmath83 @xmath1-branes are bound to it .
such a bps system may be identified with vortices on @xmath85 with magnetic flux @xmath83 . as well - known the moduli space of such vortices is the @xmath83-th symmetric product of @xmath85 which we write as @xmath269 .
there exists a classical result by macdonald on the euler - poincar ' e characteristics of symmetric products of smooth curves .
as argued in @xcite , by taking into account the fact that @xmath24-brane charges must be measured by mukai vectors , it is more natural to consider a twisted version of macdonald formula : @xmath270 a naive way to treat @xmath0-@xmath1 branes in more general settings is to extend the vortex picture relatively and to replace @xmath269 by @xmath127}$ ] .
to simplify the situation considerably we make a rather strong [ ass - integral ] all the fibers of @xmath121 are integral .
we have @xmath138}$ ] and @xmath271 . hence ,
@xmath272}_\beta)y^{\pm(d+1-g(\beta))}=\iota_\pm f_\beta(y)\,.\ ] ] from this immediately follows .
consider for instance the case of @xmath25 .
let @xmath273 $ ] denote the class of the @xmath26 . then , @xmath274}(y)=(y^{{\frac{1}{2}}}-y^{-{\frac{1}{2}}})^{-2}\,,\ ] ] and @xmath275}(y)=0 $ ] for @xmath276 .
this expression can also be obtained ( in a difficult way ! ) either by gromov - witten theory @xcite or by the topological vertex formalism @xcite . in
[ why ] we said that bubble trees in gromov - witten theory turn into @xmath1-branes ( vortices ) .
it is instructive to see how this happens on this simple example . in @xcite ,
the authors use the localization technique to evaluate the gromov - witten invariants .
this boils down to calculate the bubbling contributions at the north and south poles of the @xmath26 . on the other hand ,
the approach of the topological vertex gives a diagram consisting of one internal edge corresponding to the @xmath26 and two trivalent vertices corresponding to the north and south poles .
the torus - fixed @xmath0-@xmath1 configurations relevant to calculate @xmath277}(y)$ ] is such that a @xmath0-brane wraps singly around the @xmath26 ( so the @xmath212 young diagram is a single box ) and @xmath1-branes are localized to the north or south poles , _
i.e. _ the two vertices .
we also note that has an interpretation as a two - point function of vertex operators @xcite ( see also @xcite ) .
this matches well with the appearance of the schur polynomials in the approach of @xcite .
in the simplified setting we assume later , we will find that @xmath153 has an expansion of the form @xmath278 with @xmath279 and @xmath280 being a rational function . in fact there arise at least two natural choices of basis functions @xmath281 .
one choice is associated with gopakumar - vafa s expansion and @xmath281 takes a simple form .
the other choice has a clear geometrical meaning but @xmath281 takes a more complicated form .
in @xcite we studied an analogous case with @xmath21 replaced by a projective @xmath67 surface @xmath202 .
corresponding to @xmath121 in the present situation we considered the universal family of curves @xmath282 where @xmath283 is a smooth ( integral ) curve of genus @xmath284 on @xmath202 .
associated to @xmath285 one considers the relative compactified jacobian @xmath286 together with the abel - jacobi map @xmath287}\to \bar{\mathcal{j}}_n $ ] . with the technical assumptions there ,
all the fibers of @xmath285 are integral and both @xmath287}$ ] and @xmath288 are smooth . in @xcite
we proved an exact formula for @xmath289}_n)y^{\pm(d+1-n)}\,.\ ] ] ( the proof there rested on brill - noether theory of sheaves on @xmath67 surfaces developed by yoshioka @xcite and markman @xcite . ) in @xcite , the authors tried to develop an algorithm for calculating the numbers @xmath234 and in the process of this they also considered @xmath127}$ ] . however in their treatment it played only an auxiliary role since their point of view is that of gopakumar - vafa and not of the @xmath0-@xmath1 picture .
they applied their formalism to concrete examples and obtained satisfactory results with the caveat that one must be careful when there are reducible fibers in the family of curves ( _ e.g. _ the case of local @xmath290 ) . in the following
we will see that some of their findings can be uniformly understood once one considers the relation between the gopakumar - vafa picture and the @xmath0-@xmath1 picture .
let @xmath85 be a smooth irreducible curve of genus @xmath81 and let @xmath103 be its jacobian .
we have @xmath291 where @xmath292 is the image of the cartan generator of the lefschetz @xmath220 as before .
rather trivially , it follows that @xmath293 what implies is that one may proceed relatively : [ clconj ] @xmath294}_\beta)y^{\pm(d+1-g(\beta))}=\iota_\pm \frac{\epsilon(\beta)\lambda_\beta(y ) } { ( y^{{\frac{1}{2}}}-y^{-{\frac{1}{2}}})^{2}}\,.\ ] ] this is not obvious at all since there can be singular fibers in @xmath121 .
however , with some simplifying assumptions , this conjecture can be shown to be true .
see theorem [ mainth ] .
notice that one may take @xmath295 and @xmath296 in .
conjecture [ clconj ] implies relations between @xmath297})$ ] and @xmath234 again by proposition [ invprop ] . in particular , we note that coincides with the relation claimed in @xcite .
if @xmath298 the abel - jacobi map @xmath299}_\beta\to{\bar { \mathcal{j}}}_\beta$ ] is a fibration with fiber @xmath300 .
since @xmath301 , the relation @xmath302 is just a reflection of this projective bundle structure .
it is again instructive to see the situation for the @xmath67 surface @xmath202 as before .
there is little harm in replacing @xmath303 by the hilbert scheme of points @xmath304}:=\operatorname{hilb}^n_m$ ] .
when one views @xmath304}$ ] as a compact hyperk " ahler manifold of real dimension @xmath305 , there are a triple of k " ahler forms @xmath306 and therefore a triple of lefschetz actions @xmath307}))$ ] . according to verbitsky @xcite , @xmath308
generate an action of @xmath309 on @xmath310})$ ] .
gopakumar - vafa s @xmath221 should be identified ( with the replacement of @xmath303 by @xmath304}$ ] understood ) as a subalgebra of verbitsky s @xmath309 .
we choose one complex structure on @xmath304}$ ] thereby viewing it as an irreducible holomorphic symplectic manifold of complex dimension @xmath311 .
denote the associated k " ahler form by @xmath312 .
then @xmath313}))$ ] should be identified with the usual lefschetz action .
so the lowering operator of the @xmath220 is represented by the wedge operation of @xmath312 and the raising one by its adjoint . on the other hand ,
as we will justify shortly , @xmath314}))$ ] should be identified with the holomorphic lefschetz action first considered by fujiki @xcite .
so the lowering operator is represented by the wedge operation of a holomorphic 2-form @xmath315},\omega_{m^{[n]}}^2)$ ] and the raising one by its adjoint .
apparently an analog of @xmath232 is @xmath316})}(-1)^{h_d}y^{h_l}\,.\ ] ] ( thompson s observation ) with the convention that @xmath317 is meant for @xmath318 in the original sense of hirzebruch , we have @xmath319})\,.\ ] ] since @xmath228 acts as the multiplication of @xmath320 on @xmath321},\omega_{m^{[n]}}^p)$ ] and we have @xmath322 , we see that @xmath323 is precisely what thompson writes as @xmath324 in @xcite with @xmath2 being one of the eigenvalue of @xmath325 .
( amusingly , @xmath324 is the rozansky - witten invariant of a mapping torus @xmath326 if @xmath327 . )
so is just a consequence of his observation . to our please we showed in @xcite that @xmath289}_n)y^{\pm(d+1-n)}=\iota_\pm\frac { y^{-n}\chi_y(m^{[n]})}{(y^{{\frac{1}{2}}}-y^{-{\frac{1}{2}}})^{2}}\,.\ ] ] ( this was proved by comparing our formula for @xmath328})$ ] and the formula for @xmath329})$ ] proved earlier by g " ottsche and soergel @xcite . ) note that what corresponds to @xmath330 is @xmath331 .
as a preparation for the next section we make a digression to study hilbert schemes of points on nodal curves .
it is well - known that @xmath332}:=\operatorname{hilb}^d_c$ ] coincides with @xmath269 if @xmath85 is smooth .
this , however , is not the case when @xmath85 is singular .
suppose that @xmath333 is an integral curve of arithmetic genus @xmath81 having @xmath334 nodes and no other singularities . in this case we say that @xmath333 is @xmath334-nodal .
note that @xmath335 .
let @xmath336 be the normalization
. then @xmath337 . in general @xmath338 for the hilbert schemes we have to find the necessary correction terms to this formula , which we now turn to .
we introduce the standard notation for multinomial coefficients .
@xmath339 recall that a _ partition _ is any sequence @xmath340 of non - negative integers in non - increasing order and containing only finitely many non - zero terms .
then @xmath341 is the _ weight _ of @xmath342 and we say @xmath342 is a _ partition of @xmath83 _ if @xmath343 .
when @xmath342 is a partition of @xmath83 , we also use an alternative notation @xmath344 where @xmath345 is the multiplicity of @xmath346 in @xmath342 .
( note that @xmath347 might be zero . )
we set @xmath348 for convenience .
if @xmath344 is a partition of @xmath83 , we define @xmath349 by @xmath350 @xmath351}_{g,\delta})=\sum_{\substack{\lambda=(1^{\delta_1}\cdots d^{\delta_d})\\[2pt ] { { \lvert\lambda\rvert}=d,\ \check \delta}_1\le \delta } } a_\lambda\ , \binom{\delta}{\delta-{\check \delta}_1,\delta_2,\dots,\delta_d } \binom{\delta_1 + 1+\delta-{\check \delta}_1 - 2g}{\delta_1}\,.\ ] ] we assume @xmath352 since otherwise the assertion is trivial . let @xmath353 be the zero - dimensional subscheme of the nodes on @xmath333 so that its length is @xmath354 . for a given partition @xmath355 such that @xmath356 we partition @xmath353 into mutually disjoint union @xmath357 with @xmath358 and @xmath359 @xmath360 . in @xmath361
we consider configurations where a multiplicity @xmath346 point collides with each node of @xmath362 for each @xmath363 .
then @xmath332}_{g,\delta}$ ] differs from @xmath361 in that at each node of @xmath362 a multiplicity @xmath346 point is replaced by the punctual hilbert scheme of length @xmath346 .
after some mental exercise , @xmath364}_{g,\delta})$ ] is then found to be @xmath365 { { \lvert\lambda\rvert}=d,\ \check \delta}_1\le \delta}}\ , \sum_{\mathcal{n}=\coprod_{\ell=1}^d \mathcal{n}_\ell^\lambda}\chi\left(\bigl(c_{g,\delta}\setminus{\textstyle \coprod}_{\ell=2}^d \mathcal{n}_\ell^\lambda\bigr)^{(\delta_1)}\right)\prod_{\ell=2}^d(\chi(h_\ell)-1)^{\delta_\ell}\,,\ ] ] where @xmath366 is the punctual hilbert scheme of length @xmath346 supported at a node .
to proceed we have to calculate @xmath367 . according to ran @xcite
, @xmath366 is a rational chain @xmath368 where @xmath369 are @xmath26 s and @xmath370 are nodes .
therefore @xmath371 and we understand why the factor @xmath349 arises . next we easily see that @xmath372 finally , it should be noticed that there are @xmath373 ways to partition @xmath353 into @xmath374 .
3.2pt the above formula of @xmath364}_{g,\delta})$ ] is rather complicated and not that convenient in practice . actually there exists a neat formula for the generating function as we explain below .
however , before going to that , it will be worthwhile ( and fun ! ) to look at several examples of this formula : @xmath375 \chi(c^{[2]}_{g,\delta})&=&\binom{3+\delta-2g}{2}&+&\delta&&\\ & & \yng(1,1 ) & & \yng(2)&&\\[2 mm ] \chi(c^{[3]}_{g,\delta})&=&\binom{4+\delta-2g}{3}&+&\delta(1+\delta-2g)&+&2\delta\\ & & \yng(1,1,1 ) & & \yng(2,1)&&\yng(3 ) \end{array}\ ] ] these agree with the sample computations of @xcite .
we can give further examples : @xmath376}_{g,\delta } ) & = & \binom{5+\delta-2g}{4 } & + & \delta\binom{2+\delta-2g}{2 } & + & \binom{\delta}{2}\\[1 mm ] & & \yng(1,1,1,1 ) & & \yng(2,1,1)&&\yng(2,2)\\[2 mm ] & + & 2\delta(1+\delta-2 g ) & + & 3\delta&&\\ & & \yng(3,1 ) & & \yng(4 ) & & \end{array}\ ] ] @xmath377}_{g,\delta})&= & \binom{6+\delta-2g}{5 } & + & \delta\binom{3+\delta-2g}{3 } & + & \binom{\delta}{2}(\delta-2g)\\[1 mm ] & & \yng(1,1,1,1,1 ) & & \yng(2,1,1,1)&&\yng(2,2,1)\\[2 mm ] & + & 2\delta\binom{2+\delta-2g}{2 } & + & 2\delta(\delta-1 ) & + & 3\delta(1+\delta-2g)\\[1 mm ] & & \yng(3,1,1 ) & & \yng(3,2)&&\yng(4,1)\\[2 mm ] & + & 4\delta&&&&\\ & & \yng(5 ) & & & & \end{array}\ ] ] @xmath378}_{g,\delta})&= & \binom{7+\delta-2\,g}{6 } & + & \delta{4+\delta-2\,g\choose 4 } & + & { \delta\choose 2}{1+\delta-2\,g\choose 2 } \\[2 mm ] & & \yng(1,1,1,1,1,1 ) & & \yng(2,1,1,1,1)&&\yng(2,2,1,1)\\[3 mm ] & + & { \delta\choose 3 } & + & 2\,\delta{3+\delta-2\,g\choose 3 } & + & 2\,\delta \left ( \delta-1 \right ) \left ( \delta-2\,g \right ) \\[1 mm ] & & \yng(2,2,2 ) & & \yng(3,1,1,1)&&\yng(3,2,1)\\[3 mm ] & + & 2 ^ 2\binom{\delta}{2 } & + & 3\,\delta{2+\delta-2\,g\choose 2 } & + & 3\,\delta \left ( \delta-1 \right)\\[1 mm ] & & \yng(3,3 ) & & \yng(4,1,1)&&\yng(4,2)\\[3 mm ] & + & 4\,\delta \left ( 1+\delta-2\,g \right ) & + & 5\,\delta & & \\ & & \yng(5,1 ) & & \yng(6)&&\\[2 mm ] \end{array}\ ] ]
let us now turn to the generating function : [ nodalmacdonald ] @xmath379}_{g,\delta})y^{\pm(d+1-g)}=\iota_\pm\ , ( y^{{\frac{1}{2}}}-y^{-{\frac{1}{2}}})^{2g-2}\left(1 + ( y^{{\frac{1}{2}}}-y^{-{\frac{1}{2}}})^{-2}\right)^\delta\,.\ ] ] by eliminating the sum over @xmath380 in one obtains that @xmath381}_{g,\delta})= & \sum_{\substack{\delta_2\ge 0,\dots,\delta_d\ge 0\\[2pt ] \sum_{\ell \ge 2 } \ell \delta_\ell \le d\\ \sum_{\ell \ge 2 } \delta_\ell\le \delta } } \binom{\delta}{\sum_{\ell \ge 2 } \delta_\ell}\binom{\sum_{\ell \ge 2 } \delta_\ell}{\delta_2,\dots,\delta_d}\left(\prod_{\ell=2}^d(\ell-1)^{\delta_\ell}\right ) \\ & \qquad\quad\quad \times\binom{d-\sum_{\ell \ge 2 } \ell \delta_\ell+1+\delta- \sum_{\ell \ge 2 } \delta_\ell-2g}{d-\sum_{\ell \ge 2 } \ell \delta_\ell } \ , . \end{split}\ ] ] then consideration of the generating function leads to @xmath382}_{g,\delta})y^d&= \sum_{k=0}^\delta\,\,\sum_{\substack{\delta_2\ge 0,\delta_3\ge 0,\dots\\[2pt ] k= \sum_{\ell \ge 2 } \delta_\ell}}\binom{\delta}{k}\binom{k}{\delta_2,\delta_3,\dots}\\ & \quad\times\left(\prod_{\ell\ge 2}(\ell-1)^{\delta_\ell}\right ) y^{\sum_{\ell \ge 2 } \ell \delta_\ell}(1-y)^{2g-\delta+k-2}\ , , \end{split}\ ] ] where we understand the sequence @xmath383 contains only finitely many non - zero terms .
the right hand side can be rewritten as @xmath384 k= \sum_{\ell \ge 2 } \delta_\ell } } \binom{k}{\delta_2,\delta_3,\dots } \prod_{\ell\ge 2}\left\{(\ell-1)y^\ell(1-y)\right\}^{\delta_\ell}\ , .
\end{split}\ ] ] the multinomial theorem simplifies this as @xmath385 by summing up the geometric series one obtains @xmath386 by the binomial theorem this is equal to @xmath387 it thus follows that @xmath388}_{g,\delta})y^d=(1-y)^{2g-2}\left(1+\frac{y}{(1-y)^2}\right)^\delta\,.\ ] ] to see that this leads to is easy .
let @xmath389 be the generalized jacobian of @xmath333 and @xmath390 its compactification .
[ cjformulaprop ] @xmath391 this is ( with a correction of sign ) implicit in @xcite .
( see the discussion around eq .
( 5.8 ) of @xcite . )
we have an exact sequence of commutative algebraic groups can be obtained from one on @xmath392 by gluing the fibers over those points on @xmath392 which are identified to form nodes on @xmath333 .
for each node the ways to glue two lines are parametrized by @xmath393 . ]
@xmath394 hence @xmath390 is the product of @xmath395 and the compactification of @xmath396 .
the latter is given by @xmath397 where @xmath398 is a certain equivalence relation essentially identifying @xmath399 and @xmath400 of each @xmath26 and @xmath401 is a rational curve with a single node .
see iiib , 5 in @xcite for an exposition .
since @xmath401 is obtained by pinching a homologically non - trivial 1-cycle from an elliptic curve , we obtain @xmath402 combining this with @xmath403 we easily obtain .
the following may be viewed as an extension of .
[ nodalcl ] @xmath404}_{g,\delta})y^{\pm(d+1-g)}=\iota_\pm\ , \frac{(-1)^g\operatorname{tr}_{h^*(\overline{j(c_{g,\delta})})}(-1)^hy^h } { ( y^{{\frac{1}{2}}}-y^{-{\frac{1}{2}}})^{2}}\,.\ ] ] an immediate consequence of proposition [ nodalmacdonald ] and proposition [ cjformulaprop ] .
we now investigate how the fibration structure of @xmath121 is directly reflected to properties of @xmath153 .
we start by severi varieties were originally studied by severi ( and corrected by harris ) for curves of fixed degrees in @xmath290 .
there are numerous works on severi varieties of other surfaces .
see for instance @xcite .
here we are extending the definition for calabi - yau threefolds assuming an appropriate family of curves @xmath121 . for attempts in defining severi varieties of threefolds see @xcite and references therein .
one of the principal issues in the studies of severi varieties is whether or not they are regular .
the standard way to investigate this is to resort to a deformation theory .
there are many works for surfaces and these show that in most cases severi varieties of surfaces are regular but a care is needed when they are of general type .
see again @xcite . if @xmath408 is regular , the @xmath334 nodes can be independently smoothed and @xmath409 .
below we assume the regularity of @xmath408 for simplicity so that @xmath410 is meaningful .
@xmath412}_\beta)y^{\pm(d+1-g(\beta))}\\ & \ \ = \iota_\pm \sum_{\delta=0}^{g(\beta)}\chi(v_{\beta,\delta})\ , ( y^{{\frac{1}{2}}}-y^{-{\frac{1}{2}}})^{2g(\beta)-2}\left(1+(y^{{\frac{1}{2}}}-y^{-{\frac{1}{2}}})^{-2}\right)^\delta\ , .
\end{split}\ ] ] by our assumptions , we see ( as in @xcite ) @xmath413}_\beta)y^{\pm(d+1-g(\beta))}=\sum_{\delta=0}^{g(\beta)}\chi(v_{\beta,\delta})\ , \sum_{d=0}^\infty\chi(c^{[d]}_{g(\beta),\delta})y^{\pm(d+1-g(\beta))}\,,\ ] ] to prove it suffices to binomially expand @xmath418 on the right hand side of and then notice @xmath419 the inversion relation that leads to is well - known .
the last equality in is due to a yau - zaslow type argument @xcite .
the way gopakumar and vafa introduced @xmath234 was representation theoretic and was driven by the motivation of finding a convenient expansion basis as in .
however , to the best of my understanding , the direct geometrical characterization of @xmath234 has not yet been available .
the relation seems to give an intuitive feeling about the geometrical meaning of @xmath420 . in general
, one can produce a node on a curve by pinching a handle or gluing two points .
conversely given a @xmath334-nodal curve of arithmetic genus @xmath108 one can make a smooth curve of genus @xmath421 , where @xmath422 , by ungluing ( or partially normalizing ) @xmath423 nodes and unpinching @xmath424 nodes .
of course there are @xmath425 ways to choose such @xmath423 nodes from the entire nodes of the curve . if @xmath426 for all @xmath427 then @xmath428 .
this was the claim in @xcite .
however , such an assumption depends on @xmath423 and can not be satisfied for all @xmath423 simultaneously unless everything is empty .
in general the relation between @xmath234 and @xmath410 has to be as above . to summarize , in our extremely ideal situation we can declare any of @xmath429})\}_{d=0}^{g(\beta)}\ , , \quad \{n_\beta^h\}_{h=0}^{g(\beta)}\ , , \quad \ { \chi(v_{\beta,\delta})\}_{\delta=0}^{g(\beta)}\,,\ ] ] as a _ fundamental set of invariants_. they can be converted from one to another .
|
we discuss how the geometry of @xmath0-@xmath1 branes may be related to gromov - witten theory of calabi - yau threefolds .
|
dynamic self - organization of motile components can be observed in a wide range of length scales , from bird flocks @xcite to bacterial colonies @xcite and assemblies of motor and structural proteins @xcite .
the fascination with these phenomena has naturally inspired researchers to use a physical understanding of motility to engineer complex emergent behaviors in model systems that promise revolutionary advance in technological applications if combined with other novel biomimetic functions , such as signal processing and decision making @xcite , or replication @xcite .
symmetry - based phenomenological theories , coarse - grained or particle - based @xcite offer a guide to the rich possibilities immanent in self - driven systems , but designing a system requires a bottom - up approach .
biological components pose inevitable limitations on this task , while chemical @xcite , mechanical @xcite or externally actuated @xcite imitations appear more promising .
in addition to motility , living organisms have developed mechanisms that allow them to orient their motion in response to chemical gradients , and send signals to recruit or repel others @xcite .
can inanimate matter imitate these more complex functions ? we show that it can , and present the necessary design principles .
we consider the case of catalytic active colloids @xcite , which we now describe in brief .
recall that a colloidal particle can be driven `` phoretically '' into motion by _ externally imposed _ chemical , electrostatic or thermal gradients @xcite .
an _ active _ colloid a particle coated asymmetrically with catalyst and immersed in a _ uniform _ background of substrate generates its own chemical gradient @xcite , and thus moves autonomously in a direction determined by the polarity of the coat .
such self - phoretic particles , whose individual activity and interactions one can design , offer the opportunity to create systems with controllable , emergent collective behavior @xcite . to this end
it is essential to construct a description at a coarse - grained level , with coefficients expressed in terms of single - particle parameters .
our focus is on how the center of mass and orientation vector of an active colloid are affected by an externally imposed gradient of substrate molecules .
depending on details of geometry , activity , and mobility @xcite , an active colloid will respond to the local gradient of the substrate concentration through four distinct mechanisms .
( i ) _ chemotaxis : _ the fluid flows set up around the particle can turn its axis of orientation to align parallel or antiparallel to the local gradient ; this process has active contributions arising from the chemical reaction as well as passive ones ( ii ) _ polar run - and - tumble motion : _ the enzymatic rate depends nonlinearly on the local concentration of the substrate with a characteristic michaelis - menten form inherited from the underlying catalytic kinetics of the reactions @xcite . the combination of enhanced activity at high concentrations and randomized orientation acts to effectively populate the colloids in `` slow '' regions @xcite .
( iii ) _ apolar run - and - tumble motion : _ an active colloid can also chemotax by a net motion of its center along a gradient in a noise - averaged sense .
( iv ) _ phoretic response : _ the colloid moves along an external chemical gradient by diffusiophoresis .
a summary of the different modes is depicted in fig . 1 .
catalytic colloids consume a substrate and generate product molecules , and hence act as mobile sources and sinks of these chemicals in the solution making their concentration profile nonuniform . in a suspension of such active colloids , each individual responds via the above four mechanisms to the gradients produced by other colloids due to their activities .
the various contributions are independent of each other and their balance will be modified as we move in the space of control parameters , leading to a variety of collective behaviors . in particular , we highlight the intriguing possibility that the positional and orientational degrees of freedom could exhibit different and independent types of order depending on the parameters , as shown in figs . 2 and 3 . [
fig : scheme ] we consider a fluid medium containing a concentration @xmath0 of substrate ( s ) molecules , which upon contact with a catalyst are converted to a product p with concentration @xmath1 .
the rate of conversion @xmath2 obeys michaelis - menten kinetics @xcite , growing linearly at small @xmath3 and crossing over to saturation for sufficiently large values of @xmath3 .
our main results are as follows : * ( 1 ) * for the case of a uniform gradient of substrate , we establish the form of the angular velocity @xmath4 induced on a single catalytic colloid as a function of the spherical harmonic components of the activity @xmath5 and the mobilities @xmath6 and @xmath7 corresponding to s and p. we can therefore propose criteria for the surface patterning required to produce chemotactic and anti - chemotactic motion .
we also find the various contributions to the translational velocity @xmath8 of the colloid , arising from self - propulsion and drift due to the external gradient .
explicitly , we find @xmath9 where the definitions of the coefficients in terms of the surface properties of the colloids are given below . * ( 2 ) * we use our results for a single particle in an external gradient to construct the collective equations of motion for the number density and orientation of the colloids in a uniform medium , interacting via their effect on the substrate and product concentration fields .
the interplay between self - propulsion , phoretic drift , and alignment , driven by and mediated via chemicals , falls into two distinct regimes .
( a ) when the fuel concentration is small enough such that the catalytic activity is _ diffusion - limited _ , the chemical concentrations will be effectively screened , and the system could develop enhanced number fluctuations and clumping instabilities ( where all wavelengths above a threshold are unstable ) and patterns with a given length scale ( where the fastest - growing mode has finite wavelength ) .
( b ) at sufficiently high fuel concentrations where the catalytic activity becomes _ reaction - limited _ , the chemical fields are not screened and can mediate long - ranged interactions that could lead to a wider variety of instabilities . in particular , for the case of effectively attractive phoretic interaction we observe collapse transitions that are dissipative analogs of a jeans instability @xcite , with or without simultaneous condensation of asters . for effectively repulsive phoretic interactions
, we observe counter - intuitively that a collapsed phase with aster condensation is still possible , as are stable phases exhibiting debye - like screening , similar to electrolytes .
moreover , we find that in this regime the system could exhibit plasma - like oscillations in response to perturbations , or spontaneous , self - sustained ringing .
detailed phase diagrams inferred from our stability analysis and structure factor calculations are found in figs . 2 and 3 .
these are parameterized , via definitions in eq .
[ ab ] below , by coefficients @xmath10 describing the chemotactic response from panels ( i ) and ( ii ) of fig . 1 , and @xmath11 the phoretic response from panels ( iii ) and ( iv ) of that figure .
we now show how we obtained these results .
diffusiophoresis is the force - free , torque - free propulsion of a colloid by a solute concentration gradient @xcite . in a fluid of viscosity @xmath12 at temperature @xmath13 a species with concentration @xmath14 interacting through an effective potential @xmath15 with a particle surface with normal along the local @xmath16 axis gives rise , via the stokes equation , to a surface `` slip velocity '' @xmath17 , with the phoretic mobility @xmath18 that can have either _ sign _ depending on @xmath15 @xcite .
we are also interested here in _
self_-diffusiophoresis @xcite that occurs when @xmath19 is not imposed externally but generated by processes on the particle itself .
[ phasedia1 ] consider a single swimmer , whose mobilities and catalytic coat have the common symmetry axis @xmath20 .
when placed in a _ uniform _ substrate background , such a particle moves in a direction determined by @xmath20 if the coatings are sufficiently asymmetric @xcite .
what happens in an inhomogeneous background ?
can the flows set up by the interaction of s and p with the swimmer surface reorient its axis @xmath20 with respect to the local concentration gradient , thus imitating chemotaxis ? to answer this question ,
we solve for the concentrations @xmath3 and @xmath21 , with diffusivities @xmath22 and @xmath23 respectively .
we incorporate the catalytic chemical reaction s @xmath24 p through source and sink boundary conditions on particle fluxes normal ( @xmath25 ) to the swimmer surface : @xmath26 where @xmath27 is the probability that the enzyme at @xmath28 is bound to the substrate .
stationarity implies @xmath29 leading to the michaelis - menten @xcite expression @xmath30 for the reaction velocity per molecule .
number conservation for the products and substrates , and the assumption that @xmath3 and @xmath21 diffuse rapidly compared to the colloid so that time dependencies and advection by flow @xcite can be ignored give @xmath31 , where @xmath32 is the background substrate profile .
we thus need to solve for just one of the two concentration fields .
we work in the linear regime and @xmath33 , in which the chemical reaction influences significantly the local value of @xmath3 so that the coupled dynamics of orientation and translation can lead to oscillations , will be discussed elsewhere . ]
@xmath34 , where the profile of product @xmath21 resulting from this process is sensitive to the imposed gradient of @xmath3 , and in the limit where s diffuses rapidly so that its profile is maintained .
the resulting slip velocity , which has contributions from both the substrate and the product , leads to the linear and angular velocities @xmath35 and @xmath36 for spherical colloids . to understand the general trends in chemotactic behavior arising from simple catalytic patterns , we work with a limited number of non - zero spherical harmonic components of @xmath5 and @xmath37 . for example , taking @xmath38 for @xmath39 we find the expression for the angular velocity given in eq .
[ res_sp ] with @xmath40 where a negative ( positive ) value corresponds to chemotactic ( antichemotactic ) response . the first term on the left is the passive response to the external gradient due to a polarity in @xmath41 alone while the second term is the active contribution involving both @xmath5 and @xmath37 .
the form of eq .
[ spfom ] serves to illustrate some features that hold even without the truncated expansion in @xmath42 : if either @xmath5 or @xmath37 contain all odd or all even harmonics there is no reorientation in response to the gradient , a result which holds for spheroidal swimmers as well .
the expression for the product contribution in eq .
[ spfom ] is a sum of products of @xmath43 and @xmath44 , which can be used to design chemotactic colloids with a desired response .
lastly , regardless of the form of @xmath5 , @xmath45 for @xmath37 uniform over the sphere
. [ phasedia2 ] we also obtain the net translational velocity @xmath46 , as in eq .
[ res_sp ] where @xmath47 the three contributions to the translational velocity correspond to self - propulsion ( along @xmath48 ) , phoretic drift ( along @xmath49 ) , and an anisotropic drift that is instantaneously along @xmath48 , but leads to net motion along @xmath49 as rotational noise de - correlates @xmath48 .
the latter amounts to a contribution to run - and - tumble gradient - seeking motion , which we name apolar run - and - tumble ; see fig . 1 . to demonstrate how the chemotactic response of catalytic colloids can be designed , we have calculated @xmath50 for an example of swimmers with uniform spheroidal caps of catalytic and mobility patterns as @xmath51 and @xmath52 .
figure 4 shows @xmath50 as a function of @xmath53 for different values of @xmath54 . for @xmath55 ,
@xmath50 is antisymmetric as a function of @xmath53 . for a given @xmath54 , @xmath50 peaks near @xmath56 as the slip velocity is maximum when the position where @xmath37 is maximum coincides with the region where @xmath21 changes most rapidly , which for the given form of @xmath5 is at @xmath53
this example showcases the possibility to control the response of individual catalytic colloids , and thus their collective behaviors , by following design rules that include varying systematically their geometric features .
we now combine the individual responses of active colloids to construct a theoretical description for their collective behaviors .
catalytic swimmers of the type discussed above interact through the s and p chemical fields as well as via hydrodynamics @xcite .
we restrict our attention to their chemotactic interaction , and construct the collective behavior of many swimmers by looking at pairwise interactions .
consider , therefore , two swimmers separated by a distance @xmath57 in a uniform medium of substrate molecules .
the reaction s @xmath24 p that takes place on the surface of each swimmer modifies the @xmath3 field as seen by the others , and each is also a source for p. in the absence of a background of other swimmers , inhomogeneities in the @xmath3 and @xmath21 fields in steady state decay as @xmath58 .
each swimmer senses and responds to the magnitude and the gradient of @xmath3 through the motility and chemotaxis mechanisms outlined above .
in addition , each particle responds to the @xmath21 field produced by the reactions on the surfaces of all the particles , just as it would to any externally imposed solute gradient @xcite .
the resulting equations of motion for the position @xmath59 and orientation unit vector @xmath60 of the @xmath61th , to linear order in @xmath62 , @xmath63 , take the form @xmath64 where additional coupling constants @xmath65 are introduced to take account of the response of each colloid to a product gradient produced by the others . in eq .
[ polposeq ] thermal as well as active fluctuations are included phenomenologically via gaussian unit - strength white noise terms @xmath66 , with strengths @xmath67 and @xmath68 .
@xmath69 and @xmath70 correspond to swimmers that respond chemotactically to @xmath71 and @xmath72 respectively .
@xmath73 and @xmath74 imply attractive contributions to the interactions between the swimmers due to @xmath3 and @xmath21 respectively .
@xmath75 by definition as we choose @xmath76 to point in the direction in which a solitary swimmer moves .
the form of eq .
[ polposeq ] follows on general grounds of symmetry .
the point of our calculation is that it gives explicit expressions for the tactic and phoretic mobilities and the expressions for @xmath77 and @xmath78 .
moreover , the substrate and product fields are themselves determined by the distribution of colloid positions and orientations .
the substrate is consumed and the product is generated at the rate @xmath79 where @xmath80 is the position coordinate on the @xmath61th swimming sphere of radius @xmath81 , and the catalytic coat @xmath5 is expressed in lab - frame coordinates .
we assume the system is maintained in a steady state with mean substrate concentration @xmath82 and develop @xmath83 to leading orders in a gradient expansion to obtain a coarse - grained description .
we begin by relating the @xmath3 and @xmath21 fields to the coarse - grained density and orientation fields of the colloids , namely , @xmath84 and @xmath85 ( see the supplementary material ) .
we find @xmath86 in eq .
[ psscale ] , @xmath87 is the total number of enzymatic sites on the surface of the swimmer and @xmath88 measures the degree of polarity of the catalytic coat .
we work in the limit of @xmath89 .
linearizing eq .
[ psscale ] around a steady state with ( @xmath90 ) , we find the following results for the fourier components of the concentration at wavevector @xmath91 : @xmath92 } , \ ; p_{\mathbf{q } } = -{d_s s_\mathbf{q}}/{d_p } , \label{spsol } $ ] where we introduce _ the screening length _ @xmath93^{-1/2},\ ] ] that is a measure of the range of interactions mediated by s and p. [ omega_shape ] for @xmath94 , i.e. on the linear or _ unsaturated _ part of the mm curve , @xmath95 is finite and the interactions are therefore short - ranged . for @xmath96 ,
i.e. on the _ saturated _ part of the mm curve , @xmath97 and the product mediates an effective long - ranged interaction amongst the colloids . starting from the langevin equations in eq .
[ polposeq ] , we next construct equations of motion for @xmath98 and @xmath99 as defined above eq .
[ psscale ] . while coarse - graining , we see inevitably that the dynamical equation for @xmath100 involves higher moments of the orientational distribution function , which must be re - expressed in terms of lower moments @xcite using an appropriate closure which we discuss in the supplementary material . the s and p fields produced by inhomogeneities in density and the divergence of the polar order parameter mediate interactions between swimmers through @xmath98 and the longitudinal component @xmath101 .
the linearized dynamics in the isotropic phase closes in terms @xmath98 and @xmath102 , whose coarse - grained equations we present in the saturated limit and for wavenumbers @xmath103 in the unsaturated case .
we will see that despite the presence of a self - propelled velocity field and a density , there is an important contrast relative to models such as toner - tu @xcite : the interactions considered here offer no mechanism to promote flocking , i.e. , the global parallel alignment of @xmath100 .
we have calculated the mode structure and steady state structure factors @xmath104 and @xmath105 for both the unsaturated and the saturated cases by adding phenomenological gaussian white noise terms , conserving for @xmath98 and nonconserving for @xmath106 , to the equations of motion .
the equations of motion and their stability analysis which we now present are best shown in terms of coefficients @xmath107 , \nonumber \\ & & \hskip-1.7 cm b =
n \kappa(s_0 ) \left[\frac{1}{d_p } ( \beta_0 + \frac{\beta_1}{3 } ) + \frac{1}{d_s } ( \alpha_0 + \frac{\alpha_1}{3})\right ] , \label{ab}\end{aligned}\ ] ] which give the effective chemotactic and phoretic response to gradients respectively .
in the unsaturated case @xmath108 , for wavenumbers @xmath103 , coarse grained equations read @xmath109 { \mathbf{n}}_{l } \nonumber \\ & & + \frac{(v_1 s_0 - \rho_0 \xi_s^2 a)}{3 \rho_0 } { \mathbf{\nabla}}\rho = 0,\label{unsaturateda}\\ & & \hskip-0.7 cm [ \partial_t- ( d-\rho_0 \xi_s^2 b)\nabla^2]\rho_0^{-1 } \delta \rho + ( v_1 s_0- \epsilon \rho_0 \xi_s^2 b \nabla^2 ) { \mathbf{\nabla}}\cdot { \mathbf{n}}_l=0 , \nonumber \\ \label{unsaturatedb}\end{aligned}\ ] ] [ unsaturated ] where we use @xmath110 as the self - phoretic velocity scales linearly with the substrate concentration .
several features of eq .
[ unsaturated ] are noteworthy . from eq .
[ psscale ] , p is abundantly available or s is depleted where the density @xmath98 is high .
phoretic movement up ( down ) the gradient of @xmath111 can thus lead to a propensity of swimmers to swim up their concentration gradients and hence a change in the sign of the diffusivity in eq .
[ unsaturatedb ] through @xmath112 .
since the swimmer preferentially moves along its polar axis , @xmath113 can be viewed as a velocity field and the @xmath114 term as a pressure gradient .
we see that for large enough @xmath115 , the signs of coefficients that are analogous to bulk viscosity , and squared sound speed can change signalling an instability and possible novel condensation phenomena whose nature will be revealed only by a nonlinear treatment with appropriate noise terms .
note that @xmath10 contains two contributions : ( i ) chemotactic alignment with the local gradient in @xmath3 and @xmath21 and ( ii ) slowing down of swimmers due to increased substrate consumption as a result of a local excess of @xmath98 that depletes @xmath3 locally .
this provides a realization of the density - dependent self - propulsion velocity of @xcite . in the overdamped limit , for large @xmath10 and @xmath11
, one eigenmode with relaxation @xmath116 goes unstable with growth rate @xmath117 at small @xmath118 .
competition with stabilizing effects at larger @xmath118 will lead to a modulated growth morphology with a length scale determined by the wavenumber of peak growth @xmath119 .
the other mode , controlled by @xmath68 , remains stable for @xmath120 , i.e. , interactions do not promote flocking .
working at large @xmath68 also justifies the overdamped limit . in the parameter range where these modes are stable
, the steady - state static small-@xmath118 structure factor takes the form @xmath121 where @xmath122.\end{aligned}\ ] ] for @xmath123 , as @xmath124 , @xmath125 displays fluctuations with a correlation length @xmath126 that diverges as @xmath127 , presaging the onset of clumping ( see fig .
3 ) . still in the linearly stable regime but with @xmath128 , an analysis to order @xmath129 shows that the system has a tendency towards patterning with a characteristic length scale latexmath:[$\sim chemotactic ( @xmath10 ) and phoretic ( @xmath11 ) response to gradients .
next we consider the saturated limit @xmath97 , realized by working at saturation concentrations on the mm curve.it is useful to define @xmath131 , which plays the role of an electric field in eq .
[ saturated ] below .
equation [ psscale ] then implies @xmath132 . in this limit
the orientation and density fields satisfy we have displayed only those nonlinear terms required to stabilized a state of nonzero @xmath106 .
the complete equations may be seen in the supplementary material . ]
@xmath133 { \mathbf{n}}_l+\frac{v_0}{3 \rho_0 } { \mathbf{\nabla}}\rho + \frac{a}{3 } \mathbf{e } \nonumber \\ & & \hskip-1.2 cm - [ \frac{v_0 ^ 2}{90 d_r } - \frac{2\epsilon n \kappa_2 v_0 \rho_0}{135 d_r } ( \frac{\beta_1}{d_p}+ \frac{\alpha_1}{d_s } ) ] { \mathbf{\nabla}}{\mathbf{\nabla}}\cdot{\mathbf{n}}_l \nonumber \\ & & \hskip-1.2cm+[\frac{2 \epsilon^2 a^2 \rho_0 ^ 2 } { 15 d_r } n_l^2 - \frac{\epsilon a}{3 } \rho_0 + 2d_r ] { \mathbf{n}}_l= 0 , \label{saturateda } \\ & & \hskip-1.2 cm
( \partial_t - d\nabla^2)\rho + \rho_0 ( v_0 + \epsilon \rho_0 b ) { \mathbf{\nabla}}\cdot { \mathbf{n}}_l - \rho_0 { b}{\mathbf{\nabla}}\cdot \mathbf{e}=0 , \label{saturatedb}\end{aligned}\ ] ] where @xmath134 . the electric - field character of @xmath135 is evident in eq .
[ saturateda ] through the alignment term @xmath136 and the ohmic current @xmath137 in eq .
[ saturatedb ] .
note that a large and positive @xmath138 can destabilize the @xmath139 state leading once higher order terms are taken into account to a state of nonzero @xmath106 , i.e. a condensation of asters .
the phenomenon is related to that reported in @xcite , with the important difference in our case of long - range interactions mediated by @xmath135 , as in @xcite , with a resemblance to gravitational collapse @xcite . in the overdamped limit ,
i.e. for sufficiently large @xmath68 , the relaxation rates of the eigenmodes are @xmath140 q^2,\\\\ -2 d_r ' { + o(q^2 ) } , \end{array } \right . \ ] ] where @xmath141 represents a modified rotational diffusion , and @xmath142 is an effective control parameter for the nature of interaction between the swimmers .
equation [ saturated ] shows that in the saturated limit the effective long - ranged interaction between colloids ( as mediated by s and p ) leads to non - vanishing relaxation rates at @xmath143 for both modes , notwithstanding the conservation law governing @xmath98 . for @xmath144 ,
the swimmers interact with long - ranged repulsive interactions .
the structure factor @xmath145 , as it is a ratio of the strength of fluctuations and the wavenumber independent relaxation rate , which is reminiscent of suppression of charge density fluctuations in electrolytes . including terms of higher order in @xmath118 yields a density structure factor with a peak at @xmath146 , which is characteristic of micro - phase separation ( white region in fig .
we see from eq .
[ saturated ] that for @xmath147 and large @xmath148 the isotropic state with uniform density is linearly unstable for small wavenumber @xmath118 , including @xmath143 .
this effect is a dissipative analog of the gravitational jeans instability @xcite , and is a consequence of the long - ranged attractive interaction ( yellow ) .
related behavior has been predicted for thermophoretic colloids @xcite .
letting @xmath149 by increasing @xmath10 and keeping @xmath147 brings the system out of the overdamped region where the relaxation of @xmath113 slows down and it behaves like a velocity field .
this behavior where the system resembles a gravitational system conserving momentum and displays an instability formally equivalent to the standard hydrodynamic jeans instability @xcite ( magenta ) .
modes with wavenumbers larger than a crossover scale given by a competition between the interaction strength @xmath150 and a squared sound speed equivalent @xmath151 for our system are oscillatory , whereas modes with smaller wavenumbers are too `` massive '' and collapse . on further tuning the parameters to approach @xmath152 , notwithstanding the value of @xmath150 ,
one anticipates an instability towards a spontaneously oscillating state ( green ) . restricting our attention to the stable case
, we find a structure factor for @xmath106 with a correlation length @xmath153 that grows as @xmath148 decreases , indicating strong fluctuations towards aster formation . for small @xmath148 and @xmath144 the response shows `` plasma oscillations '' @xcite with frequency @xmath154 . for @xmath152 ( dark green ) , the system can also develop spontaneous oscillations , or ringing .
a colloid patterned with catalyst and immersed in a maintained reactant medium is a minimal nonequilibrium particle , displaying directed motion and related behaviours ruled out at thermal equilibrium .
we have determined theoretically the nature of patterning that will cause such an active colloid to reorient along and move up or down a gradient of chemical reactant , thus delineating the principles for the design of chemotactic self - phoretic particles .
coarse - graining the resulting langevin equations for the position and polar axis of one particle , we discover the dynamics of the density and polar order parameter of a collection .
the interplay of chemotaxis and phoresis leads to clumping and patterning at low reactant concentration ; at high concentration , the slow decay of diffusing reactants and products yields analogues of electrostatic and gravitational phenomena debye screening , microphase separation , plasma oscillations and gravitational collapse .
the interactions promote aster formation , not a flocking transition , and the instabilities mediated by the long - range diffusion fields have a character distinct from those the generic instability driven by the velocity field in stokesian active liquid crystals .
we look forward to experimental tests and , eventually , practical application of our predictions .
rg and sr acknowledge hfsp grant rgp0061/2013 and a j c bose fellowship respectively , and both thank the isaac newton institute for mathematical sciences where a part of this work was completed . ss thanks tcis for hospitality . 10 cavagna a , giardina i ( 2014 ) bird flocks as condensed matter . _ annu rev cond matt phys _ 5 .
ben - jacob e , cohen i , levine h ( 2000 ) cooperative self - organization of microorganisms . _
adv phys _ 49:395 .
zhang hp , beer a , florin el , swinney hl ( 2010 ) collective motion and density fluctuations in bacterial colonies . _
pnas _ 107:13626?13630 .
huber f _ et al . _
( 2013 ) emergent complexity of the cytoskeleton : from single filaments to tissue .
_ adv phys _ 62:1 .
schweitzer f ( 2007 ) brownian agents and active particles : collective dynamics in the natural and social sciences ( springer series in synergetics ) , springer the swarms project webpage .
www.swarms.org .
wang t _ et al . _
( 2011 ) self - replication of information - bearing nanoscale patterns .
_ nature _ 478 : 225 - 228 .
toner j , tu y ( 1995 ) long - range order in a two - dimensional dynamical xy model : how birds fly together .
_ phys rev lett _ 75:4326 - 4329 .
ramaswamy s ( 2010 ) the mechanics and statistics of active matter .
_ annual reviews _ 1 : 323 - 345 .
marchetti mc et al . ( 2013 ) hydrodynamics of soft active matter . _
rev mod phys _ 85 : 1143 - 1189 .
grgoire g , chat h ( 2004 ) onset of collective and cohesive motion .
_ phys rev lett _ 92:025702 .
vicsek t , zafeiris a ( 2012 ) collective motion .
_ phys rep _ 517:71 .
lau awc , lubensky tc ( 2009 ) fluctuating hydrodynamics and microrheology of a dilute suspension of swimming bacteria .
e _ 80:011917 .
bialk j , lwen h , speck t ( 2013 ) microscopic theory for the phase separation of self - propelled repulsive disks .
_ epl _ 103:30008 .
thutupalli s , seemann r , herminghaus s ( 2011 ) simple model squirmers with tunable velocity .
yamada d , hondou t , sano m ( 2003 ) coherent dynamics of an asymmetric particle in a vertically vibrating bed .
e _ 67 : 040301 .
dreyfus r _
( 2005 ) microscopic artificial swimmers .
_ nature _ 437 : 862 .
keller ef , segel la ( 1971 ) travelling bands of chemotactic bacteria : a theoretical analysis .
_ j theor biol _ 30:235 .
paxton wf , et al .
( 2004 ) catalytic nanomotors : autonomous movement of striped nanorods .
_ j am chem soc _ 126:13424 .
golestanian r , liverpool tb , ajdari a ( 2005 ) propulsion of a molecular machine by asymmetric distribution of reaction products .
_ phys rev lett _ 94:220801 .
howse jr , et al . (
2007 ) self - motile colloidal particles : from directed propulsion to random walk .
_ phys rev lett _ 99:048102 .
jiang , h - r , yoshinaga n , sano m ( 2010 ) active motion of a janus particle by self - thermophoresis in a defocused laser beam .
_ phys rev lett _ 105:268302 .
anderson jl ( 1989 ) colloid transport by interfacial forces .
_ annu rev fluid mech _ 21:61 .
rckner g , kapral r ( 2007 ) chemically powered nanodimers .
_ phys rev lett _ 98:150603 .
golestanian r , liverpool tb , ajdari a ( 2007 ) designing phoretic micro- and nano - swimmers .
_ new j phys _ 9:126 .
mitchell p ( 1956 ) hypothetical thermokinetic and electrokinetic mechanisms of locomotion in micro - organisms .
_ 25:32 - 34 .
mitchell p ( 1972 ) self - electrophoretic locomotion in microorganisms : bacterial flagella as giant ionophores .
_ febs lett .
_ 28:1 - 5 . lammert pe , prost j , bruinsma r ( 1996 ) ion drive for vesicles and cells .
_ j. theor .
_ 178:387 - 391 .
hong y , blackman nmk , kopp nd , sen a , velegol d ( 2007 ) chemotaxis of non - biological nanorods .
_ 99:178103 .
ibele me , mallouk te , sen a ( 2009 ) schooling behavior of light - powered autonomous micromotors in water .
_ angew . chem .
_ 48:3308 - 3312 .
golestanian r ( 2009 ) anomalous diffusion of symmetric and asymmetric active colloids .
_ 102 : 188305 . ibele me _
( 2010 ) emergent , collective oscillations of self - mobile particles and patterned surfaces under redox conditions .
_ acs nano _ 8:4845 - 4851 .
kagan d , balasubramanian s , wang j ( 2011 ) chemically triggered swarming of gold microparticles .
_ 50:503 - 506 .
theurkauff i , cottin - bizonne c , palacci j , ybert c , bocquet l ( 2012 ) dynamic clustering in active colloidal suspensions with chemical signaling . _ phys rev lett _ 108:268303 .
palacci j , sacanna s , steinberg ap , pine dj , chaikin pm ( 2013 ) living crystals of light - activated colloidal surfers .
_ science _ 339:936 - 940 .
duan w , liu r , sen a ( 2013 ) transition between collective behaviors of micromotors in response to different stimuli .
_ 135:1280 - 1283 .
baraban l , harazim sm , sanchez s , schmidt og ( 2013 ) chemotactic behavior of catalytic motors in microfluidic channels .
ed . _ 52:5552 - 5556 .
ebbens s , tu mh , howse jr , golestanian r ( 2012 ) size dependence of the propulsion velocity for catalytic janus - sphere swimmers .
_ phys rev e _ 85:020401 .
cates m e ( 2012 ) diffusive transport without detailed balance in motile bacteria : does microbiology need statistical physics ? .
_ rep prog phys _ 75:042601 nelson pc ( 2007 ) biological physics .
wh freeman fair mc , anderson jl ( 1989 ) electrophoresis of nonuniformly charged ellipsoidal particles . _
j. colloid interface sci .
_ 127 : 388 .
alexander gp and liu ap ( 2011 ) self - diffusiophoresis in the advection dominated regime .
arxiv:1107.3851 .
jeans j ( 1929 ) astronomy and cosmogony , 2nd edition .
cambridge univ . press ,
cambridge .
lushi e , goldstein re , shelley mj ( 2012 ) collective chemotactic dynamics in the presence of self - generated fluid flows .
e _ 86:040902 . zttl a , stark h ( 2013 ) hydrodynamics determines collective motion and phase behavior of active colloids .
arxiv:1309.4352 .
lauga e , thomas rp ( 2009 ) the hydrodynamics of swimming microorganisms .
_ 72:096601 .
bertin e , droz m , gregoire g ( 2006 ) boltzmann and hydrodynamic description for self - propelled particles .
_ phy rev e _ 74 : 022101 .
saintillan d , shelley mj ( 2008 ) instabilities , pattern formation , and mixing in active suspensions _ phys fluids _ 20 : 123304 .
fily y , marchetti mc ( 2012 ) athermal phase separation of self - propelled particles with no alignment .
_ phys rev lett _ 108 : 235702 .
gowrishankar k , rao m ( 2012 ) nonequilibrium phase transitions in active contractile polar filaments .
arxiv:1201.3938v1 .
golestanian r ( 2012 ) collective behavior of thermally active colloids .
_ phys rev lett _ 108:038303 .
tonks l , langmuir i ( 1929 ) oscillations in ionized gases .
rev . 33:195 - 210 .
we begin by obtaining an expression for the sources @xmath155 for generation of p and consumption of s in terms of the coarse - grained density and orientation fields , @xmath156 , @xmath157 , of the colloids . @xmath83
simply sums over all the enzymatic sites on the surfaces of all catalytic colloids . setting @xmath158 ( see fig . [ q_schem ] ) , @xmath159 as the infinitesimal solid angle and taylor expanding upto @xmath160 we get @xmath161 \nonumber \\ & & \times \left[\sigma_0 + \sigma_1 \mathbf{\hat{x}}_\alpha \cdot \mathbf{\hat{n}}_\alpha+ \sigma_2 \left(\mathbf{\hat{x}}_\alpha \cdot \mathbf{\hat{n}}_\alpha\right)^2\right ] \nonumber \\ & = & \kappa(s ) 4 \pi r^2 \left [ \sigma_0 \rho - \frac{1}{3 } \sigma_2 \rho- \frac{r \sigma_1 } { 3 } \mathbf{\nabla } \cdot \mathbf{n}\right ] \nonumber \\ & = & \kappa(s ) n \left [ \rho - \frac{\sigma_2}{3 \sigma_0 } \rho- \frac{r \sigma_1 } { 3 \sigma_0 } \mathbf{\nabla } \cdot \mathbf{n}\right ] , \label{monodi}\end{aligned}\ ] ] where @xmath162 .
the sources are obtained using @xmath163 [ q_schem ] .,width=288 ]
next , starting from the langevin equations , we construct equations of motion for @xmath164 and @xmath165 .
we construct the joint probability density @xmath166 for the position and orientation ; @xmath167 @xmath166 is used to derive the fokker planck equation for the joint distribution of particle position and polarity ; @xmath168 - d \nabla^2 p \nonumber \\ & & + \left(\mathbf{\hat{w } } \times \mathbf{\nabla}_{w}\right ) \cdot \left[p\left(\phi_0 \left(\mathbf{\hat{w } } \times \mathbf{\nabla}s\right)\times \mathbf{\hat{w } } + \omega_0 \left(\mathbf{\hat{w } } \times \mathbf{\nabla } p\right)\times \mathbf{\hat{w}}\right)\right ] - d_r \left(\mathbf{\hat{w } } \times \mathbf{\nabla_{w}}\right)^2 p = 0 \end{aligned}\ ] ] the equation for @xmath169 is given by taking the zeroth moment of the above equations @xmath170
-d \nabla^2 \rho = 0 , \label{rhoeq}\end{aligned}\ ] ] where @xmath67 is the diffusivity of the swimmers and @xmath171 is the nematic order parameter .
the equation for the polar order parameter @xmath172 is the first moment of @xmath173 - d\nabla^2 \mathbf{w } \nonumber \\ & + & \phi_0 \left[{\textsf{\textbf{q } } } \cdot \mathbf{\nabla}s - \frac{2}{3 } \rho \mathbf{\nabla}s \right ] + \omega_0 \left[{\textsf{\textbf{q } } } \cdot \mathbf{\nabla } p - \frac{2}{3 } \rho \mathbf{\nabla } p \right ] + 2d_r \mathbf{w } = 0 , \label{eqn } \end{aligned}\ ] ] where @xmath174 . while coarse - graining , we see inevitably that the dynamical equation for @xmath175 involves higher moments of the orientational distribution function which must be re - expressed in terms of lower moments @xcite .
no process involving apolar bundling of particles is considered here , so we do not consider strict nematic order .
the occurrence of more than one factor of @xmath176 can then lead only to higher powers of @xmath175 , combined with isotropic terms .
we obtain a hierarchy of equation which have to be truncated using a closure condition .
the condition often used is to assume that @xmath177 relaxes fast so that @xmath178 , i.e. @xmath177 is slaved to @xmath98 and @xmath179 . in this scheme of closure
we systematically ignore terms higher than @xmath180 so that it holds close to the order disorder transition .
deep in the ordered state the closure scheme has to be modified keeping in mind that @xmath181 , where @xmath182 is the ordering direction and @xmath183 is magnitude , so that one has to look at fluctuations about this reference state .
we will now write down the equation of motion for the nematic order parameter and obtain an expression for @xmath177 in terms of @xmath98 and @xmath113 .
the expression for @xmath177 is substituted in the equations for @xmath98 and @xmath113 to obtain coarse grained equations of these variables in terms of @xmath98 and @xmath113 and involving the diffusion field @xmath3 .
next we work in two different approximations to obtain expressions for @xmath3 in terms of @xmath98 and @xmath113 to obtain effective equations for @xmath98 and @xmath113 .
the nematic order parameter @xmath177 satisfies the equation @xmath184 - \alpha_0 \mathbf{\nabla } \cdot \left ( \mathbf{\nabla}s { \textsf{\textbf{q } } } \right ) - \alpha_1 \mathbf{\nabla } \cdot \left({\textsf{\textbf{q}}}^4 \cdot \mathbf{\nabla } s - \frac{{\textsf{\textbf{i}}}}{3 } { \textsf{\textbf{q}}}^{\prime } \cdot \mathbf{\nabla }
s\right ) + \beta_0 \mathbf{\nabla } \cdot \left(\mathbf{\nabla}p { \textsf{\textbf{q}}}\right ) \nonumber \\ & & + \beta_1 \mathbf{\nabla } \cdot \left({\textsf{\textbf{q}}}^4 \cdot \mathbf{\nabla } p - \frac{{\textsf{\textbf{i}}}}{3 } { \textsf{\textbf{q}}}^\prime \cdot \mathbf{\nabla } p\right ) - d \nabla^2 { \textsf{\textbf{q } } } + 2\phi_0 \left({\textsf{\textbf{q}}}^{3 } \cdot \mathbf{\nabla } s - \mathbf{w}\mathbf{\nabla}s\right ) + 2\omega_0 \left({\textsf{\textbf{q}}}^{3 } \cdot \mathbf{\nabla}p - \mathbf{w}\mathbf{\nabla}p\right ) \nonumber \\ & & + 6 d_r { \textsf{\textbf{q}}}= 0 , \label{eqq}\end{aligned}\ ] ] where @xmath185 and @xmath186 .
+ the traceless parts of higher moments like @xmath187 and @xmath188 are assumed to be negligibly small , so that @xmath189 using and @xmath190 , and defining for any two vectors @xmath191 and @xmath192 , @xmath193 and keeping terms upto the lowest order in gradients we have @xmath194^{ts } , \label{qsol } \end{aligned}\ ] ] where @xmath195 we also define @xmath196,\nonumber \\ a & = & n \kappa(s ) [ \frac{\phi_0}{d_s } - \frac{\omega_0}{d_p } + \delta \frac{v_0(s_0)}{2 d_s}],\end{aligned}\ ] ] where @xmath197 . in this regime ,
the phoretic velocity is independent of the substrate concentration so that @xmath198 and we substitute @xmath199 and @xmath200 . substituting @xmath177 from into and using @xmath201 we have the equation of motion for the polar order parameter @xmath202 \nonumber \\ & & + \frac{c_2 ^ 2}{10 d_r } { \mathbf{n}}|{\mathbf{\nabla}}s|^2 + \frac{c_2 ^ 2}{30d_r}\left({\mathbf{n}}\cdot { \mathbf{\nabla}}s\right){\mathbf{\nabla}}s \nonumber \\ & & + \frac{c_2c_3}{270d_r } \left [ 2 \rho \left(\nabla^2s\right){\mathbf{\nabla}}s - 6 \rho\left({\mathbf{\nabla}}{\mathbf{\nabla}}s\right)\cdot { \mathbf{\nabla}}s - 3 { \mathbf{\nabla}}\rho |{\mathbf{\nabla}}s|^2 - { \mathbf{\nabla}}s \left({\mathbf{\nabla}}s \cdot { \mathbf{\nabla}}\rho\right ) \right]=0\end{aligned}\ ] ] in the saturated limit we have @xmath203 and @xmath204 .
we have defined the curl free or longitudinal component of @xmath113 expressed in the fourier space as @xmath205 . in what follows
we use @xmath206 and define @xmath207 . the effective equation for @xmath175
is then @xmath208 + [ { \mathbf{n}}{\mathbf{n}}\mathbf{e } ] + [ ( \mathbf{e}-\epsilon { \mathbf{n}}_l ) { \mathbf{\nabla}}\rho ] + [ { \mathbf{n}}{\mathbf{\nabla}}{\mathbf{n}}]=0,\end{aligned}\ ] ] where the square brackets denote a linear combination of all possible contractions of the enclosed terms to obtain a vector . the equation is asymptotically exact as @xmath97 .
we will now discuss the significance of various terms and provide expressions for the functions introduced above . as the swimmer moves with an average speed @xmath209 along its symmetry axis
, the polar order parameter is equivalent to a velocity field . with this interpretation @xmath98
becomes a pressure like term where corrections to the squared sound speed @xmath210 is given by @xmath211 similarly @xmath212 is a generalised pressure term depending on the other fields as @xmath213 . \nonumber\end{aligned}\ ] ] the presence of the swimmer creates spatial variation in the s and p. this is encoded in the fields @xmath135 and @xmath106 of which the first enters as an external orienting field while the second enters the dynamics as a spontaneous symmetry breaking effect where the linear term @xmath214 can change sign producing @xmath106 condensation .
there are higher order stabilising terms with coefficient @xmath215 which restrain the growth of @xmath216 .
the linear terms coefficients are modified by the corrections @xmath217 , \nonumber \\ & & f_4 = \frac{a c_3}{135 d_r } \rho \delta
\rho + \frac { \epsilon^2 a^2 } { 30 d_r } |{\mathbf{n}}_l|^2 + \frac{a c_3}{270 d_r } { \mathbf{\nabla}}\rho \cdot\left(\mathbf{e } - \epsilon { \mathbf{n}}_l\right),\end{aligned}\ ] ] which can destabilise the condensed state .
note that the terms that dictate the nature of the aster condensation are all proportional to the polarity of the catalytic coat as they are sensitive to the polar nature of the diffusion field .
the advective nonlinearities have the same structure as in the toner - tu model @xcite given that @xmath113 is like a velocity that can carry other fields .
however note the absence of terms of the form @xmath218 ; this is because the interaction in the two cases are very different and in this particular closure that holds close to the transition the average value of @xmath113 and thus the coefficients of advective terms are proportional to the interaction strength which in this case is @xmath219 .
bertin et .
@xcite modelled the toner - tu like interaction as an actual binary collision which tends to align the particles . in this system where interactions are mediated by the diffusion fields alone ,
the interaction is mediated by the @xmath106 term as discussed above .
@xmath220 & \equiv & \epsilon \left(\frac { c_3}{5 } - \frac{a c_3 \epsilon}{135 d_r } \rho + \frac{11 a v_0 } { 90 d_r}\right ) \left({\mathbf{\nabla}}\cdot { \mathbf{n}}\right ) { \mathbf{n}}_l + \epsilon\left(\frac{a v_0 } { 10 d_r } + \frac{c_3}{5 } \right ) \left({\mathbf{n}}\cdot { \mathbf{\nabla}}\right ) { \mathbf{n}}_l \nonumber \\ & & \hskip-1 cm - \epsilon \left(\frac{a v_0}{10 d_r } + c_1\right ) { \mathbf{n}}\left({\mathbf{\nabla}}\cdot { \mathbf{n}}\right ) + \epsilon \left(\frac{a v_0}{15 d_r } + c_1\right ) \left({\mathbf{n}}_l \cdot { \mathbf{\nabla}}\right ) { \mathbf{n}}- \epsilon \frac{a v_0 } { 30 d_r } { \mathbf{n}}_l \times \left({\mathbf{\nabla}}\times { \mathbf{n}}\right ) .
\nonumber\end{aligned}\ ] ] the other nonlinearities are the terms coupling @xmath113 and @xmath135 which to lowest order are @xmath221 & \equiv & \left(\frac { c_3}{5 } - \frac{a c_3 \epsilon}{135 d_r } \rho + \frac{11 a v_0 } { 90 d_r } \right ) \left({\mathbf{\nabla}}\cdot { \mathbf{n}}\right)\mathbf{e } + \left(c_1 + \frac{a v_0 } { 15 d_r}\right)\left(\mathbf{e } \cdot { \mathbf{\nabla}}\right ) { \mathbf{n}}\nonumber \\ & & + \left(\frac { c_3}{5 } + \frac{a v_0 } { 10 d_r}\right)\left({\mathbf{n}}\cdot { \mathbf{\nabla}}\right)\mathbf{e}+ \frac{a v_0}{30 d_r } \mathbf{e } \times \left({\mathbf{\nabla}}\times { \mathbf{n}}\right ) , \nonumber\end{aligned}\ ] ] and higher order terms @xmath222 & \equiv & - \frac{\epsilon a^2 } { 45 d_r } { \mathbf{n}}\left({\mathbf{n}}_l \cdot \mathbf{e}\right ) - \frac{\epsilon a^2 } { 30 d_r } { \mathbf{n}}_l \left({\mathbf{n}}\cdot \mathbf{e}\right ) + \frac{a^2}{30 d_r } \left({\mathbf{n}}\cdot \mathbf{e}\right)\mathbf{e}. \nonumber\end{aligned}\ ] ] the rest of the nonlinearities couple @xmath114 and @xmath223 : @xmath224= \frac{v_0 c_3}{90 d_r } \left[(\mathbf{e}-\epsilon { \mathbf{n}}_l ) \cdot { \mathbf{\nabla}}\right]{\mathbf{\nabla}}\rho + \frac{v_0 c_3}{30 d_r } \left({\mathbf{\nabla}}\rho \cdot { \mathbf{\nabla}}\right)\left(\mathbf{e } - \epsilon { \mathbf{n}}_l\right ) + \frac{v_0 c_3}{90 d_r } \left(\mathbf{e } - \epsilon { \mathbf{n}}_l\right)\nabla^2 \rho .
\nonumber\end{aligned}\ ] ] we linearise to obtain @xmath225 defining @xmath113 by @xmath226 ; and the transverse component of @xmath113 as @xmath227 we get @xmath228 { \mathbf{n}}_l + \left ( \frac{4\epsilon^2 a^2 } { 30 d_r } + 2 d_r- \frac{2 a \epsilon \rho_0}{3}\right ) { \mathbf{n}}_l\nonumber \\ & & + \rho_0^{-1}\left(\frac{v_0}{3 } - \frac{2 c_3 v_0 \rho_0}{135 d_r}\right ) { \mathbf{\nabla}}\rho + \frac{2 a}{3 } \mathbf{e}=0 \label{linnl}\\ & & \left[\partial_t - \left(d+ \frac{v_0 ^ 2}{30 d_r}\right ) \nabla^2 + \left(\frac{2\epsilon c_3 v_0
\rho_0}{135 d_r } - \frac{v_0 ^ 2}{90 d_r}\right ) { \mathbf{\nabla}}{\mathbf{\nabla}}\cdot\right ] { \mathbf{n}}_t + \left ( \frac{\epsilon^2 a^2 \rho_0 ^ 2}{10 d_r } { \mathbf{n}}_l^2 + 2 d_r\right ) { \mathbf{n}}_t = 0\end{aligned}\ ] ] note that for @xmath229 , @xmath230 stays in the overdamped region and is a fast degree of freedom that we throw away .
fourier transforming eq .
we get @xmath231 i\mathbf{q } \cdot { \mathbf{n}}_{l\mathbf{q } } - \frac{v_0}{3 \rho_0 } q^2 \rho_{\mathbf{q } } + \frac{2 a}{3 } \rho_{l\mathbf{q } } = 0,\end{aligned}\ ] ] substituting the closed form for @xmath177 from eq . into eq .
we get the equation for @xmath98 @xmath232 = 0\end{aligned}\ ] ] substituting for @xmath233 and using @xmath234 we have @xmath235 \nonumber \\ & & + \epsilon { \mathbf{\nabla}}\cdot { \mathbf{n}}_l \left(b \rho + \frac{\epsilon c_3 ^ 2}{135 d_r } \rho \delta
\rho + \frac{\epsilon^2 c_3 a}{30 d_r } w_l^2 \right ) -\frac{c_3}{15d_r}{\mathbf{\nabla}}\cdot \mathcal{j } = 0,\end{aligned}\ ] ] where @xmath236 and @xmath237 is the nonlinear part of the current : @xmath238 eq is linearised and written in terms of @xmath113 : @xmath239 fourier transforming we get @xmath240 in the linear part of the mm curve where @xmath241 , @xmath242 varies linearly with substrate concentration so that we can define the self phoretic velocity as @xmath243 .
the equation for @xmath113 in the unsaturated limit is @xmath244\left({\mathbf{\nabla}}\cdot { \mathbf{n}}\right ) { \mathbf{\nabla}}s - \left[\frac{\left(c_2 + 5 { v_1}\right){v_1}}{30d_r}s + c_1\right]\left({\mathbf{\nabla}}s \cdot { \mathbf{\nabla}}\right){\mathbf{n}}- \left[\frac{\left(3c_2 + { v_1}\right){v_1}}{30d_r}s + \frac{c_3}{5}\right]\left({\mathbf{n}}\cdot { \mathbf{\nabla}}\right ) { \mathbf{\nabla}}s \nonumber \\ & & - \left[\frac{\left(3c_2+{v_1}\right){v_1}}{30d_r } s+c_1\right ] { \mathbf{n}}\nabla^2 s + \left[\frac{\left(3c_2 + { v_1}\right){v_1}}{45 d_r}s-\frac{c_3}{5}\right]{\mathbf{\nabla}}\left({\mathbf{n}}\cdot { \mathbf{\nabla}}s\right ) -\left[\frac{\left(2 { v_1}- c_2\right){v_1}}{30 d_r}\right]{\mathbf{\nabla}}s \times \left({\mathbf{\nabla}}\times { \mathbf{n}}\right ) \nonumber \\ & & + \frac{2c_3 { v_1}}{135d_r } s \nabla^2\left({\mathbf{\nabla}}s\right ) -
\frac{\left({v_1}- c_2\right)\left(3c_2 + { v_1}\right)}{90 d_r } \left [ { \mathbf{n}}|{\mathbf{\nabla}}s|^2 + 3 { \mathbf{\nabla}}s \left({\mathbf{n}}\cdot { \mathbf{\nabla}}s\right ) \right ] \nonumber \\ & & + \frac{c_3\left({v_1}-c_2\right)}{90d_r } \left[\rho { \mathbf{\nabla}}|{\mathbf{\nabla}}s|^2 - \frac{2}{3 } \rho \nabla^2 s { \mathbf{\nabla}}s \right ] + \frac{\left(c_2+{v_1}\right ) c_3}{90 d_r}\left [ \frac{1}{3 } { \mathbf{\nabla}}s \left({\mathbf{\nabla}}s \cdot { \mathbf{\nabla}}\rho\right ) + { \mathbf{\nabla}}\rho |{\mathbf{\nabla}}s|^2 \right ] \nonumber \\ & & + \frac{c_3 { v_1}s}{270 d_r } \left [ 9 \left({\mathbf{\nabla}}\rho \cdot { \mathbf{\nabla}}\right ) { \mathbf{\nabla}}s - 2 \left(\nabla^2 s\right ) { \mathbf{\nabla}}\rho + \left(\nabla^2 \rho\right ) { \mathbf{\nabla}}s + 3 \left({\mathbf{\nabla}}s \cdot { \mathbf{\nabla}}\right){\mathbf{\nabla}}\rho - 2 { \mathbf{\nabla}}\left({\mathbf{\nabla}}\rho \cdot { \mathbf{\nabla}}s\right ) \right]= 0\end{aligned}\ ] ] to obtain the effective equation for @xmath113 we substitute @xmath245 ; in the highly unsaturated limit we have @xmath246 for @xmath247 .
keeping terms upto second order in gradients we get @xmath248 \nonumber \\ & & - \frac{\xi_s^4 ( 3 { v_1}n \kappa - a d_s)(3a d_s - { v_1}n \kappa ) } { 360 d_r d_s^2 } \left [ 3 { \mathbf{n}}|{\mathbf{\nabla}}\rho|^2 + { \mathbf{\nabla}}\rho({\mathbf{n}}\cdot { \mathbf{\nabla}}\rho)\right ] \nonumber \\ & & + \xi_s^2 [ { \mathbf{\nabla}}\rho { \mathbf{\nabla}}{\mathbf{n}}]= 0.\end{aligned}\ ] ] the pressure is modified due to two separate contributions : ( 1 ) because of chemotactic alignment of swimmers with the local chemical gradient through @xmath249 and ( 2 ) due to the change in self phoretic velocity with change in background substrate concentration through the term @xmath250 .
the advective nonlinearities are @xmath251 = \nonumber \\ & & - \left[\frac{\left(13 { v_1}n \kappa - 11 a d_s \right){v_1}s_0}{180d_r d_s } - \frac{n \kappa c_3}{5 d_s}\right]\left({\mathbf{\nabla}}\cdot { \mathbf{n}}\right ) { \mathbf{\nabla}}\rho + \left[\frac{\left(a d_s + 9{v_1}n \kappa \right){v_1}s_0}{60d_r d_s}+ \frac{n \kappa c_1}{d_s } \right]\left({\mathbf{\nabla}}\rho \cdot { \mathbf{\nabla}}\right){\mathbf{n}}\nonumber \\ & & + \left[\frac{\left(3a d_s -{v_1}n \kappa \right){v_1}s_0}{60d_r d_s } + \frac{c_3 n \kappa}{5 d_s}\right]\left({\mathbf{n}}\cdot { \mathbf{\nabla}}\right ) { \mathbf{\nabla}}\rho + \left[\frac{\left(3a d_s -{v_1}n \kappa \right){v_1}s_0}{60d_r d_s } + \frac{n \kappa c_1}{d_s } \right ] { \mathbf{n}}\nabla^2 \rho \nonumber \\ & & + \left[\frac{c_3 n \kappa}{5 d_s } - \frac{\left ( 3a d_s -{v_1}n \kappa \right){v_1}s_0 } { 90 d_r d_s } \right ] { \mathbf{\nabla}}\left({\mathbf{n}}\cdot { \mathbf{\nabla}}\rho\right ) + \frac{\left(5 { v_1}n \kappa - a d_s\right){v_1}}{60 d_r d_s}{\mathbf{\nabla}}\rho \times \left({\mathbf{\nabla}}\times { \mathbf{n}}\right).\end{aligned}\ ] ] note that the advective nonlinearities given in eqs . and are very different in the two regimes because in the unsaturated limit @xmath106 acts like an ordering field while in the other the local chemical gradient is given by @xmath252 .
the linearised equation for @xmath113 is @xmath253 the equation for @xmath106 and @xmath230 are @xmath254 { \mathbf{n}}_l + \frac{\epsilon \rho_0 \xi_s^2 a}{3 } { \mathbf{\nabla}}\left({\mathbf{\nabla}}\cdot { \mathbf{n}}_l\right ) + \left[\frac{{v_1}s_0}{3 \rho_0 } - \frac { \xi_s^2 a } { 3 } \right ] { \mathbf{\nabla}}\rho = 0 , \nonumber \\ & & \left[\partial_t + 2d_r - \left(d+\frac{{v_1}^2 s_0 ^ 2}{30 d_r}\right)\nabla^2 \right]{\mathbf{n}}_t = 0\end{aligned}\ ] ] fourier transforming the equation for @xmath106 get @xmath255 i\mathbf{q } \cdot { \mathbf{n}}_{l\mathbf{q } } - q^2\frac{\left[{{v_1}s_0 } - \rho_0 a \xi_s^2 \right]}{3 \rho_0 } \rho_{\mathbf{q } } - q^2 \frac{\epsilon \rho_0 a}{3 } i\mathbf{q } \cdot { \mathbf{n}}_{l\mathbf{q}}= 0 , \end{aligned}\ ] ] the equation for @xmath98 is @xmath256 where @xmath237 is the nonlinear part of the current . substituting for @xmath257 , the effective equation for @xmath98 upto @xmath258 is @xmath259 \nonumber \\ & & + { \mathbf{\nabla}}\cdot { \mathbf{n}}\left[{v_1}s_0 - \frac{{v_1}n \kappa \xi_s^2}{d_s } \left(\delta \rho - \epsilon { \mathbf{\nabla}}\cdot { \mathbf{n}}\right ) \right ] + \frac{n
\kappa c_3}{45 d_r d_s}{\mathbf{\nabla}}\cdot \mathcal{j } = 0,\end{aligned}\ ] ] with the current @xmath260 - { v_1}\left(s_0 - \frac{n \xi_s^2 \kappa}{d_s } \delta \rho\right)\left({\mathbf{\nabla}}\rho \cdot { \mathbf{\nabla}}\right){\mathbf{n}}\nonumber \\ & & - \frac{{v_1}}{2 } \left(s_0 - \frac{n \xi_s^2 \kappa } { d_s } \delta \rho\right ) { \mathbf{\nabla}}\rho \times\left({\mathbf{\nabla}}\times { \mathbf{n}}\right ) - \frac{{v_1}}{3 } \left(s_0 - \frac{n \xi_s^2 \kappa}{d_s } \delta \rho\right ) \left({\mathbf{\nabla}}\cdot { \mathbf{n}}\right){\mathbf{\nabla}}\rho = 0\end{aligned}\ ] ] the linearised equation for @xmath98 written in terms of @xmath113 is @xmath261 \rho_0^{-1 }
\rho + \left({v_1}- \epsilon \xi_s^2 b \rho_0
\nabla^2\right ) { \mathbf{\nabla}}\cdot { \mathbf{n}}= 0\end{aligned}\ ] ]
in the saturated limit the dynamical matrix takes the form @xmath262,\end{aligned}\ ] ] where @xmath263 .
the mode structure is @xmath140 q^2,\\\\ -2 ( d_r-\frac{\rho_0 b}{2 } ) - [ 2 d+\frac{2 v_0 ^ 2}{45 d_r } - \frac{2\epsilon n \kappa_2 v_0 \rho_0}{135 d_r } ( \frac{\beta_1}{d_p}+ \frac{\alpha_1}{d_s } ) ] q^2 , \end{array } \right .
\nonumber \\ & & \hskip-1.2cm\mbox{where }
g = { 2\rho_0 b d_r } + \frac{1}{3}\rho_0 a v_0.\end{aligned}\ ] ] note that @xmath150 gives the strength and sign of effective long ranged interaction between the centres of mass of two swimmers taking into account both phoretic and chemotactic response of swimmers to s and p gradients .
we see in that the main novelty of the saturated limit is that it leads to a non - vanishing relaxation rate at @xmath143 for both modes , notwithstanding the conservation law governing @xmath98 . for @xmath264 , there is a wide separation in time scale between the two modes ; the first mode whose corresponding eigenvector is mostly @xmath98 relaxes slowly compared to the other which is predominantly @xmath113 .
this makes it possible to make the dynamics of @xmath113 slaved to @xmath98 and obtain a meaningful effective equation for the latter .
we will show that the @xmath98 dynamics is equivalent to that of an electrolyte or a gravitational system in a passive frictional medium for @xmath144 or @xmath147 respectively .
@xmath144 implies a repulsive interaction between the swimmers ; @xmath98 and @xmath265 in this case are analogous to the charge density and ohmic conductivity respectively in an electrolyte .
remember that the static structure factor is simply the strength of fluctuations divided by the rate at which they decay ; for @xmath98 the fluctuations are number conserving so that they go as @xmath266 while the relaxation is anomalously fast for wavelengths larger than a screening length due to the non local nature of the coulomb - like interactions . to obtain the structure factors
requires , of course , that we add phenomenological , gaussian , spatiotemporally white noise terms , conserving for @xmath98 and nonconserving for @xmath106 of strength @xmath267 and @xmath268 , to the equations of motion .
the full expressions for the structure factors are @xmath269 q^2 + ... } \nonumber \\ s_n & = & \frac{2 b^2 \mathcal{p}/d_r^2 + \mathcal{n } a^2/9d_r^2 } { -g/2d_r + \left[\frac{v_0(v_0 + \epsilon \rho_0 b)}{3 d_r } + g d/4 d_r ' d_r \right ] q^2 + ... } \end{aligned}\ ] ] neglecting terms that are small when @xmath68 is large . there is screening behaviour where fluctuations at length scales larger than @xmath270}{-g}}\end{aligned}\ ] ] are suppressed in an essentially wavenumber independent way .
including terms of higher order in @xmath118 indicated in the ellipsis in yields a density structure factor with a peak at @xmath146 , characteristic of microphase separation .
fluctuations in @xmath113 are suppressed and @xmath271 just decays . for @xmath147 , the mode structure in shows that there is an instability for @xmath147 so that the structure factors in have to be interpreted differently now . for wavenumbers larger than @xmath272 the fluctuations die off whereas for wavenumbers smaller than @xmath272 the system self gravitates and collapses ; this is a dissipative jeans instability .
@xmath273 also shows a peak as density gradients produced by @xmath98 tends to align @xmath113 .
these effects are reminiscent of behaviour predicted for a collection of thermophoretic colloids @xcite .
increasing @xmath10 or @xmath11 and thus slowing down the relaxation of @xmath106 takes one out of the overdamped regime .
the general expressions for the eigenfrequencies are @xmath274 which can now become oscillatory . in the underdamped limit @xmath113
behaves like a velocity field and the systems resembles a system conserving momentum with some subtle differences discussed in context of the advective nonlinearities in the appendix . for @xmath147 and @xmath149
we see in that a mode structure similar to the known hydrodynamic jeans instability is obtained with corrections due to disspation .
modes with wavenumber larger than @xmath275 @xmath276 are oscillatory and the ones smaller than @xmath275 are too ` massive ' and collapse .
note that the crossover wavenumber is given by a competition between the interaction strength @xmath150 and sound speed square equivalent @xmath277 for our system .
for @xmath278 we now have a electrolyte that shows plasma oscillations with typical frequency @xmath279 as @xmath280 . if in addition @xmath152 it is possible that the oscillations could become spontaneous .
the static structure factors on the stable side are @xmath281 ( d_r ' + d q^2 ) } , \nonumber \\ s_n & = & \frac{\mathcal{n } } { d_r'+d q^2 } + \frac { b^2 \mathcal{p } + \mathcal{n } a^2/9 } { \left [ \frac{2v_0\left(v_0 + \epsilon \rho_0 b \right)}{3 } q^2 - g\right]}.\end{aligned}\ ] ] we find a structure factor for @xmath106 with correlation length @xmath282 indicating strong fluctuations towards aster formation . on approaching @xmath149 , from the stable side the peak values of @xmath125 and @xmath273 increase as seen in . for @xmath147 and @xmath283
we have a jeans instability with a very strongly correlated hydrodynamic field ; however extensive study of this interesting regime is beyond the scope of our present work .
|
the creation of synthetic systems that emulate the defining properties of living matter , such as motility , gradient - sensing , signalling and replication , is a grand challenge of biomimetics .
such imitations of life crucially contain active components that transform chemical energy into directed motion .
these artificial realizations of motility point in the direction of a new paradigm in engineering , through the design of emergent behavior by manipulating properties at the scale of the individual components .
catalytic colloidal swimmers are a particularly promising example of such systems . here
we present a comprehensive theoretical description of gradient - sensing of an individual swimmer , leading controllably to chemotactic or anti - chemotactic behavior , and use it to construct a framework for studying their collective behavior .
we find that both the positional and the orientational degrees of freedom of the active colloids can exhibit condensation , signalling formation of clusters and asters .
the kinetics of catalysis introduces a natural control parameter for the range of the interaction mediated by the diffusing chemical species . for various regimes in parameter space in the long - ranged limit
our system displays precise analogs to gravitational collapse , plasma oscillations and electrostatic screening .
we present prescriptions for how to tune the surface properties of the colloids during fabrication to achieve each type of behavior .
|
the procedure of determining parton distributions by so - called global fits to data , mainly structure functions , is long established @xcite-@xcite
. however , it is a rather more recent development to try to determine the errors on these distributions at the same time .
this has come about for a number of reasons .
firstly , the sheer amount of data ( full references in @xcite ) sensitive to various parton distributions , and the precision of this data , has become such that an accurate determination of all parton distributions is possible ( with some problems only in difficult to reach regions of phase space , e.g. @xmath0 very near to 1 ) .
secondly , the understanding of the experimental errors on this data has reached a new level of sophistication , with the systematic errors being understood far better in terms of their separate sources and correlations .
lastly , the theoretical understanding at nlo in @xmath1 has improved so that subtleties due to e.g. heavy quarks are now understood .
there are many issues in the determination of errors on parton distributions , and a discussion of these may be found in @xcite
. however , one of the main outstanding problems , and the focus of a discussion session at this meeting , is the manner in which one determines precisely the size of the errors .
the quality of the fit to a set of data is generally presented in terms of the @xmath2 .
taking fully into account the correlated systematic errors this may be calculated by the covariance matrix method , i.e. the covariance matrix is constructed as @xmath3 where @xmath4 runs over each source of correlated systematic error and @xmath5 are the correlation coefficients .
the @xmath2 is then defined by @xmath6 where @xmath7 is the number of data points , @xmath8 is the measurement and @xmath9 is the theoretical prediction depending on parton input parameters @xmath10 .
alternatively , one can incorporate the systematic errors into the theory prediction @xmath11 where @xmath12 is the one - sigma correlated error for point @xmath13 from source @xmath4 .
the @xmath2 is then defined by @xmath14 where the second term constrains the values of @xmath15 , assuming that the correlated systematic errors are gaussian distributed .
this is identical to the correlation matrix definition of @xmath2 if the errors are small . in many cases
the statistical and systematic errors are simply added in quadrature , either for simplicity , because this has much the same result as the above procedures , or due to lack of information on the correlations of systematic errors .
( the sophisticated treatment is found to be essential for tevatron jets where correlated systematic errors dominate . )
all groups performing global fits use some combination of the above ways to calculate @xmath2 , and minimizing with respect to the defined @xmath2 completely determines the parton distributions . in general , the quality of the total fit is reasonably good , e.g. for cteq6 one obtains @xmath16 and for mrst2001 @xmath17 .
, as a function of @xmath18,width=340 ] in principle , the one-@xmath19 error on some parameter in the fit , e.g. the value of @xmath18 or one of the parameters describing the input parton distributions , can be determined by allowing the value of @xmath2 to vary one unit from its minimum . in practice , this results in unrealistically small errors .
this is shown in fig . 1 , which shows the variation of the @xmath2 with @xmath18 for the total fit , and for individual sets within the global fit for the mrst partons .
it is also demonstrated in an alternative manner in fig . 2 , where for the cteq6 fit the extraction of @xmath18 obtained by allowing @xmath2 to increase by one unit for each data set within the fit is presented @xcite . clearly
the size of the errors and the scatter of the points are unrealistic . obtained by cteq from the quality of fit to individual data sets within the global fit.,width=283 ]
this failure of the @xmath20 rule may have various sources .
firstly , it is clear from examination that not all of the data sets are consistent with each other , i.e. the errors have been underestimated in some manner . thus fitting the data
simultaneously leads to a certain degree of inconsistency .
one could keep only those sets which are clearly consistent with each other ( within the context of nlo qcd ) @xcite .
however , this leads to a severe restriction in the amount of data used , and the partons in some regions would lose important constraints .
( due to the correlation between different regions of @xmath0 via sum rules and because evolution equations for partons involve convolutions over a range of @xmath0 , problems in one range of @xmath0 impact on the partons over the whole range . )
a second problem is that certain assumptions have to be made when performing a global fit .
these include the cuts made on data , data sets included in the fit , the parameterization for input parton sets , the form of strange sea , the precise heavy flavour prescription , the definition of the nlo strong coupling constant , the starting scale of evolution _ etc .
_ many of these can lead to variations considerably greater than those obtained from @xmath20 .
finally , we do not have a perfect theory
nlo - in-@xmath1 qcd has many corrections .
obviously there are higher orders ( nnlo ) , which are not that small in themselves in qcd , but there are other problems .
in particular at large and small @xmath0 there are large logarithms associated with higher orders , i.e. at small @xmath0 there are terms of the type @xmath21 , and at large @xmath0 terms like @xmath22 . these are not fully understood . also at low @xmath23
there are higher twist , i.e. @xmath24 corrections that are again not well understood .
hence , in fitting data one parameter may be artificially well constrained at some incorrect value in some region of parameter space by the necessity to account for missing corrections to the theory .
this can then influence other regions in the manner explained above , e.g. an artificially large gluon in one range of @xmath0 can lead to it being too small in other regions because of the momentum sum rule constraint .
this has led to various methods to determine errors by allowing a wider variation than the canonical @xmath20 .
cteq actually allow @xmath25 by various considerations , including error limits of individual experiments @xcite .
an example is the determination of @xmath18 as shown in fig . 3
, where the error bars give the @xmath26 confidence limit for each experiment ( the size of which varies considerably between data sets ) , and the two lines give the band of @xmath18 allowed by @xmath25 for the global fit @xcite .
arguably this is too conservative , and cteq regard it as more than a one-@xmath19 error .
mrst use @xmath27 as an approximate one-@xmath19 error instead , based on the essentially subjective criterion of judging when the fit is going uncomfortably wrong in some region @xcite . obtained by cteq from the quality of fit to individual data sets within global fit.,width=283 ]
a more precisely defined method is to use an alternative formulation for @xmath2 . in the offset method
the fit is obtained by minimizing @xmath28 i.e. the best fit and parameters @xmath29 are obtained from only uncorrelated errors .
the quality of the fit can then be estimated by adding errors in quadrature .
the systematic errors on @xmath30 are determined by letting each @xmath31 and adding the deviations in quadrature .
this is essentially the same as , and in practice achieved by calculating 2 hessian matrices @xmath32 and defining covariance matrices @xmath33 this method was used in early h1 fits @xcite and in zeus fits @xcite .
much the same method is used for polarized data in @xcite , as far as the information on correlated errors allows . strictly speaking ,
the method is not optimum , and it leads to a much bigger uncertainty than the standard method for @xmath20 @xcite .
however , it may be suitable in practice @xcite , and zeus estimate the same results could be obtained using the standard method and @xmath34 .
the results of the various approaches may be summarized in a selection of the extracted values of @xmath18 below @xcite .
@xmath35 the values obtained are consistent , and the errors not too dissimilar given the wide variation in @xmath36 used .
this is largely because each group has chosen a method which gives a reasonable and believable error .
h1 actually use the standard definition , obtaining a small error , but are only able to do this because they use a much smaller data set than the other groups , i.e. their own data plus some bcdms data which are fully consistent .
if mrst and cteq were to use the same approach their error would be tiny ( and their extracted value of @xmath18 some standard deviations away from that of h1 , e.g. mrst2001 would obtain @xmath37 ) .
in determining errors for and from parton distributions one encounters a dilemma .
one could fit to a relatively small subset of the available data with a very conservative set of cuts so that data are self consistent and theory is truly reliable .
however , this would lead to a poor determination of partons in certain regions and would give no idea how well the theory is working overall .
alternatively , one can try to constrain the partons in as wide a region of parameter space as possible , accepting that the standard rules for error determination will break down .
the way in which one can obtain sensible estimates of the real errors in the latter case is a subject very much open for discussion .
|
a discussion is presented of the manner in which uncertainties in parton distributions and related quantities are determined .
one of the central problems is the criteria used to judge what variation of the parameters describing a set of partons is acceptable within the context of a global fit .
various ways of addressing this question are outlined .
|
over the last few decades , quantum spin ladders have been actively studied mainly in connection with spin - liquid behaviour , quantum critical points and superconductivity under hole doping of some cuprates ( see ref . [ ] for a review ) . in particular , the frustrated spin-1/2 heisenberg two - leg ladder exhibits a striking dimerized ground state [ ] and a low - temperature magnetization process with an intermediate plateau and magnetization jumps [ ] .
another challenging topic of current research interest consists of the theoretical investigation of related models such as the quantum spin-1 heisenberg two - leg ladder [ ] .
the main goal of the present work is to find the exact ground states of a simpler spin-1 ising - heisenberg ladder and to contrast them with the respective ground states of the pure quantum spin-1 heisenberg ladder .
note that the former model is analytically tractable using the procedure developed in refs .
[ ] and it brings insight into the relevant behaviour of the latter not fully integrable model .
consider first the frustrated spin-1 ising - heisenberg ladder with the heisenberg intra - rung interaction and the unique ising interaction along the legs and diagonals .
the total hamiltonian of the investigated model is given by @xmath0 , \label{ham}\end{aligned}\ ] ] where @xmath1 denotes spatial components of the spin-1 operator , the former suffix @xmath2 = 1 or 2 enumerates the leg and the latter suffix specifies a lattice position within a given leg .
the coupling constant @xmath3 denotes the isotropic heisenberg intra - rung interaction , the parameter @xmath4 determines the ising interaction along the legs and diagonals , @xmath5 is an external magnetic field . for further convenience ,
let us introduce the spin operator @xmath6 , which corresponds to the total spin angular momentum of the @xmath7th rung .
it can be easily proved that the operators @xmath8 and @xmath9 commute with the hamiltonian ( [ ham ] ) , i.e. @xmath10 = [ \hat{\cal h } , \hat{\textbf{t}}_{i}^z ] = 0 $ ] , which means that the total spin of a rung and its @xmath11 component represent conserved quantities with well defined quantum numbers .
the complete energy spectrum of the frustrated spin-1 ising - heisenberg ladder then readily follows from the relation @xmath12 which depends just on the quantum numbers @xmath13 and @xmath14 determining the eigenvalues of the total spin of the @xmath7th rung and its @xmath11th spatial projection , respectively . using this procedure
, the spin-1 ising - heisenberg two - leg ladder has been rigorously mapped to some classical chain of composite spins and accordingly , we can readily find all available ground states by looking for the lowest - energy state of eq .
( [ diag ] ) .
next , we will also consider the frustrated spin-1 heisenberg two - leg ladder defined by the hamiltonian @xmath15 , \label{hamh}\end{aligned}\ ] ] which represents the pure quantum analogue of the frustrated spin-1 ising - heisenberg ladder discussed previously . taking advantage of the definition for the total spin of each rung ,
the hamiltonian ( [ hamh ] ) of frustrated spin-1 heisenberg two - leg ladder can be rewritten into the form @xmath16 according to eq .
( [ diagh ] ) , the frustrated spin-1 heisenberg ladder can be rigorously decomposed into the direct sum of quantum spin chains with spin 0 , 1 or 2 at each site .
the ground state of such a system can be shown to be either a homogeneous chain , with the same spin at all sites , or a chain with alternating spins on every other site . comparing the energy of the different chains , obtained either analytically or using dmrg simulations , the exact ground - state phase diagram of the frustrated spin-1 heisenberg ladder can be constructed [ ] .
the constructed ground - state phase diagrams of the frustrated spin-1 ising - heisenberg and heisenberg ladders are depicted in fig.[fig1]a and fig.[fig1]b , respectively .
the ground states of the spin-1 ising - heisenberg ladder can be discerned according to the @xmath11th projection of the total spin on two consecutive rungs @xmath17 $ ] , because @xmath18 holds for all available ground states .
the quantum ground states [ 0,0 ] , [ 0,1 ] , [ 1,1 ] , [ 2 , 1 ] and [ 2,0 ] represent six different phases , whereas @xmath19 implies a formation of two singlets on the @xmath7th rung , @xmath20 entails only one singlet and @xmath21 denotes fully polarized rungs without singlets .
besides , two classical ground states [ 2,2 ] and [ 2,-2 ] are pertinent to the ferromagnetic and antiferromagnetic ordering .
the magnetization normalized with respect to its saturation equals zero for [ 0,0 ] and [ 2,-2 ] , one - quarter for [ 0,1 ] and [ 2,-1 ] , one - half for [ 1,1 ] and [ 2,0 ] , three - quarters for [ 2,1 ] and unity for [ 2,2 ] .
altogether , it can be concluded that the frustrated spin-1 ising - heisenberg ladder always exhibits a stepwise magnetization curve , which involves intermediate plateaux at one - quarter , one - half and three - quarters of the saturation magnetization that are however of different origin .
it is quite clear from fig.[fig1]b that the ground - state phase diagram of the pure quantum heisenberg ladder exactly coincides with that of the ising - heisenberg ladder just for sufficiently weak inter - rung interactions @xmath22 . a relatively good agreement between both phase diagrams
is still observed in the parameter space @xmath23 , where the gapless phase [ 2 ] with a continuously varying magnetization is present between the intermediate plateaux instead of direct magnetization jumps .
the gapless phase [ 2 ] corresponds to the luttinger - liquid phase of the effective spin-2 quantum heisenberg chain .
finally , the gapped haldane phase of the effective spin-2 quantum heisenberg chain emerges for @xmath24 at sufficiently low fields . in conclusion
, we have rigorously found the ground states of the frustrated spin-1 ising - heisenberg and heisenberg ladders in a magnetic field .
it has been demonstrated that the ising - heisenberg ladder always exhibits a stepwise magnetization curve with three different intermediate plateaus , while the same quantum ground states can be identified in the pure quantum heisenberg ladder provided the intra - rung coupling is sufficiently large
. 8 m.t .
batchelor , x.w .
guan , n. oelkers , z. tsuboi , adv . phys . * 56 * , 465 ( 2007 ) . m.p .
gelfand , phys .
b * 43 * , 8644 ( 1991 )
. a. honecker , f. mila , m. troyer , eur .
j. b * 15 * , 227 ( 2000 ) .
v. ravi chandra , n. surendan , phys .
b * 74 * , 024421 ( 2006 ) .
f. michaud , t. coletta , s.r .
manmana , j .- d .
picon , f. mila , phys .
b * 81 * , 014407 ( 2010 ) .
barry , m. meisel , phys .
b * 58 * , 3129 ( 1998 ) .
barry , j.d .
cohen , m. meisel , int . j. mod .
b * 23 * , 1981 ( 2009 ) .
o. rojas , j. streka , m.l .
lyra , phys . lett .
a * 377 * , 920 ( 2013 ) .
|
ground states of the frustrated spin-1 ising - heisenberg two - leg ladder with heisenberg intra - rung coupling and only ising interaction along legs and diagonals are rigorously found by taking advantage of local conservation of the total spin on each rung .
the constructed ground - state phase diagram of the frustrated spin-1 ising - heisenberg ladder is then compared with the analogous phase diagram of the fully quantum spin-1 heisenberg two - leg ladder obtained by density matrix renormalization group ( dmrg ) calculations .
it is demonstrated that both investigated spin models exhibit quite similar magnetization scenarios , which involve intermediate plateaux at one - quarter , one - half and three - quarters of the saturation magnetization .
|
equilibrium equations in the form of complementarity conditions often appear as constraints in optimization problems .
problems of this type are commonly referred to as mathematical programs with complementarity constraints ( mpccs ) , and arise in many engineering and economic applications ; see the survey @xcite and the monographs @xcite for further references .
the growing collections of test problems @xcite indicate that this an important area .
mpccs can be expressed in general as [ mpec ] ll & f(x , y , w , z ) + & x x + & f(x , y , w , z ) = 0 + & 0 y w 0 , where @xmath0 is a polyhedral set , @xmath1 and @xmath2 are twice continuously differentiable functions , and @xmath3 .
the complementarity constraint @xmath4 means that either a component of @xmath5 is zero , or the corresponding component of @xmath6 is zero , which implies that @xmath7 .
many approaches exist for solving mpccs .
these include branch - and - bound methods @xcite , implicit nonsmooth approaches @xcite , piecewise sqp methods @xcite , perturbation and penalization approaches @xcite analyzed in @xcite , and the penalty interior - point algorithm ( pipa ) @xcite , whose convergence properties are the subject of this note .
recently , there has also been renewed interest in solving ( [ mpec ] ) with standard nonlinear programming ( nlp ) techniques by replacing the complementarity constraint by @xmath8 this approach had been regarded as numerically unsafe because ( [ mpec ] ) violates a constraint qualification at any feasible point @xcite .
however , numerical @xcite and theoretical evidence @xcite suggests that sequential quadratic programming ( sqp ) solvers are able to solve large classes of mpccs .
a similar analysis is being extended to interior - point methods ( ipms ) @xcite .
this note is organized as follows .
the next section briefly describes pipa and presents its convergence result .
section [ example ] presents a small counterexample that shows that pipa can fail to converge for certain admissible choices of parameters .
finally , section [ conclusion ] discusses the example and suggests a remedy that avoids this problem .
pipa ( * ? ? ?
* chapter 6.1 ) solves mpccs of the form ( [ mpec ] ) .
it has been studied in @xcite and has been applied to a number of real applications in finance @xcite , target classification @xcite , and electricity markets @xcite .
pipa generates a sequence of iterates @xmath9 that are strictly interior in @xmath10 , @xmath11 .
the iterates are computed by solving a quadratic direction - finding problem . in that sense
, pipa combines aspects of interior - point and sqp methods .
we use @xmath12 to denote the step or displacement computed by pipa .
subscripts such as @xmath13 refer the part of @xmath12 corresponding to the @xmath14-variables ; superscripts are used to denote iterates or evaluation of functions at a particular point , for example , @xmath15 .
diagonal matrices are denoted by @xmath16 and @xmath17 and @xmath18 .
the algorithm solves the following direction - finding problem at every iteration : [ qp - dir ] ll & f^k^t d + d_x^t q^k d_x + & x^k + d_x x + & f^k + f^k^t d = 0 + & y^k d_w + w^k d_y = - y^k w^k + e + & d_x ^2_2 c ( f^k + y^k^tw^k ) , where @xmath19 and @xmath20 are parameters and @xmath21 approximates second - order information in @xmath1 and @xmath2 .
note that ( [ qp - dir ] ) has a quadratic trust - region - type constraint .
if this constraint is replaced by the @xmath22 norm , then ( [ qp - dir ] ) becomes a quadratic program ( qp ) .
note also that the linearization of the complementarity constraint @xmath23 is relaxed by @xmath24 , giving pipa an interior - point flavor .
note that pipa is not strictly an interior - point algorithm , as it remains interior only with respect to the complementary variables @xmath25 but not with respect to the other constraints .
we will show that the final trust - region type constraint is the source of the problems affecting pipa . after determining a search direction
, pipa performs a backtracking line - search to ensure that @xmath26 remain positive and close to the central path .
this is achieved by finding the root @xmath27 of [ e - g ] g_k ( ) = ( 1 - ) + ( _ 1 i m d^k_y_i d^k_w_i - ) .
global convergence is enforced by further reducing the step length to yield sufficient reduction in a quadratic penalty function [ e - p ] p_(x , y , w , z ) = f(x , y , w , z ) + ( f(x , y , w , z ) ^2 + y^t w ) , where the penalty parameter @xmath28 is chosen to be the smallest @xmath29 such that the step ensures sufficient model decrease , [ e - moddecr ] f^k^t d - ^p
( 1- ) y^k^t w^k < - ^p ( 1- ) y^k^t w^k < - y^k^t w^k .
note that the penalty function @xmath30 mixes a quadratic penalty for the nonlinear equation @xmath2 with an @xmath31 penalty for complementarity ( since @xmath32 , it follows that @xmath33 ) .
further details of the algorithm can be found in @xcite .
pipa can be summarized as follows .
* penalty interior - point algorithm ( pipa ) * = = = 1 .
_ initialization : _ + choose parameters @xmath19 , @xmath34 , + choose a starting point @xmath35 such that @xmath36 suitably centered , + choose a penalty parameter @xmath37 and set @xmath38 . +
* repeat * + 2 . solve problem ( [ qp - dir ] ) for a trial step @xmath39 .
+ 3 . find a step size @xmath40 to : + 3.1 ensure centrality and positivity of @xmath41 by finding the root of @xmath42 in ( [ e - g ] ) + or setting @xmath43 if this root does not exist or is greater than 1 .
+ 3.2 ensure sufficient reduction in the quadratic penalty function @xmath44 + in ( [ e - p ] ) by performing an armijo - search on @xmath30 .
+ let @xmath45 be the step size determined in 3.1 and 3.2 .
@xmath46 , @xmath47 .
+ * until * @xmath48 this presentation of pipa is less sophisticated than @xcite , which allows for instance @xmath49 to vary with the iteration . convergence of pipa is established in @xcite under the following two assumptions .
* [ sc ] * strict complementarity of the solution , namely @xmath50 . * [ ns ] * nonsingularity of the matrix @xmath51 .\ ] ] a sufficient condition of * [ ns ] * is that the jacobian of @xmath2 satisfies the mixed @xmath52 property .
the following theorem summarizes the convergence results in @xcite . in the next section ,
we present a small example for which pipa converges to a non - stationary point , contradicting theorem [ pipa - conv ] .
in this section , we show that pipa may fail to converge to a stationary point .
this result contradicts the convergence results of @xcite .
this section examines the behavior of pipa applied to the following small example : [ example ] ll & x + w + & -1 x 1 + & -1 + x + y = 0 + & 0 y w 0 .
the solution of ( [ example ] ) is @xmath53 , which satisfies the assumptions * [ sc ] * and * [ ns ] * , since @xmath54 and @xmath55 \ ; = \ ; \left [ \begin{array}{ll } 1 & 0 \\ 0 & 2 \end{array } \right]\ ] ] is nonsingular .
next , we show that this example generates a limit point that is not stationary .
specifically , we perform a standard iteration of the algorithm and bound the limit away from stationarity . the following lemma summarizes our result
. * proof . *
the starting point satisfies the linear equation , which implies that they are satisfied for all subsequent iterations .
the theorem follows by induction . clearly , the assertions hold for the starting point
. now assume that ( [ e - ind-1 ] ) to ( [ e - ind-5 ] ) hold for @xmath56 , and show that they also hold for @xmath57 .
the direction - finding problem for this example can be simplified as follows : [ ex - qp ] ll & d_x + d_w + & -1 x^k + d_x 1
+ & d_x + d_y = 0 + & w^k d_y + y^k d_w = ( - 1 ) y^k w^k + & - d_x , where @xmath58 for simplicity .
however , the example remains valid for certain positive semi - definite bounded @xmath21 .
first we show that the optimal basis @xmath59 to the lp ( [ ex - qp ] ) is given by @xmath60^{-1 } \ ; = \ ; \left [ \begin{array}{ccc } 0 & 1 & -\frac{w^k}{y^k } \\ 0 & 0 & \frac{1}{y^k } \\ 1 & -1 & \frac{w^k}{y^k } \end{array } \right ] .\ ] ] from this , multipliers can be computed as @xmath61 \ ; \left ( \begin{array}{c } 1 \\ 0 \\ 1 \end{array } \right ) \ ; = \ ; \left ( \begin{array}{c } -\frac{w^k}{y^k } \\ \frac{1}{y^k } \\ 1 + \frac{w^k}{y^k } \end{array } \right).\ ] ] optimality follows , since @xmath62 for the only inequality constraint .
the step @xmath12 can also be computed as @xmath63 where @xmath64 is the right - hand side of the active constraints .
note that for any positive semi - definite @xmath21 , this solution remains unchanged as long as @xmath62 .
the contribution of the hessian to the multipliers can be estimated as @xmath65 which remains positive for @xmath66 or @xmath67 sufficiently small .
thus , the same conclusions apply even if a positive definite @xmath21 is used .
next , we show that ( [ e - ind-3 ] ) follows by applying the step @xmath68 , using @xmath69 and induction : x^k+1 & = & x^k + ^k d^k_x + & & x^k - + & & x^k - _ l=0^k ( ) ^l + & = & ( ) .
[ e - xk ] it is easy to show that the expression in the last line is greater than @xmath70 , which implies that the `` choice '' of active set in the lp ( [ ex - qp ] ) was correct , and the lower bound on @xmath14 is never active during the iteration .
next we show that ( [ e - ind-1 ] ) follows by applying the step in the @xmath5 component , @xmath71 next consider ( [ e - ind-5 ] ) , and observe that @xmath69 follows trivially . to obtain the lower bound , observe that the root of @xmath42 in ( [ e - g ] ) is given by the following expression , where the superscripts @xmath57 have been omitted for the sake of simplicity @xmath72 from the fact that @xmath73 , @xmath74 , induction , and @xmath75 , the step size can be bounded below by its leading terms : @xmath76 it remains to show that this lower bound on @xmath27 also passes the second line - search criterion . the penalty update rule in (
* ( 6.1.24 ) ) finds the smallest integer @xmath77 such that @xmath78 holds , where we have used the fact that @xmath79 for all @xmath57 .
since @xmath80 and @xmath81 , this is always satisfied for @xmath82 , and the penalty parameter is never increased .
finally , this step size also satisfies the conditions for the armijo search in step 3.2 .
the actual reduction clearly satisfies @xmath83 since @xmath84 , it follows that @xmath85 which implies the sufficient reduction condition , @xmath86 since @xmath87 and @xmath88 .
next consider ( [ e - ind-2 ] ) .
since @xmath89 , it follows that [ e - wnew ] w^k+1 = w^k + ( - w^k - ) w^k - .
thus , it is possible to bound the new complementarity violation after the step by @xmath90 obtaining ( [ e - ind-4 ] ) .
this concludes the proof .
@xmath91 by taking the limit ( @xmath92 ) in ( [ e - xk ] ) , it follows , that all control iterates satisfy the bound [ e - xlim ] x^k+1 - 0.4828 . similarly , an upper bound can be derived for the state iterates as @xmath93 which shows that [ e - ylim ] y^k+1 1 + 1.4828 .
these bounds can also be verified numerically .
example ( [ example ] ) and pipa have been implemented in ampl @xcite by using the `` looping extension '' , which allows the convenient implementation of algorithms in ampl .
the sequence that is generated is given in table [ t - res ] and confirms the bounds given above .
the columns headed `` ared '' and `` pred '' in table [ t - res ] refer to the actual and predicted reduction .
their implications are discussed below .
.iterates , actual and predicted reduction of ampl implementation of pipa [ cols=">,>,>,>,>,>",options="header " , ] thus , it follows that all iterates remain in a compact set , namely , @xmath94 and the sequence @xmath95 has a limit point @xmath96 . from ( [ e - wnew ] ) it follows that @xmath97 so that @xmath98 and @xmath99
. therefore , pipa converges to a limit point @xmath96 such that @xmath100 this limit point satisfies complementarity and the linear constraints for problem ( [ example ] ) ( since @xmath101 satisfies the linear constraints ) .
moreover , this limit point also satisfies the assumptions of theorem [ pipa - conv ] .
assumption * [ sc ] * is satisfied because @xmath102 , and assumption * [ ns ] * holds because @xmath103 \ ; = \ ; \left [ \begin{array}{ll } 1 & 0 \\ 0 & y^{\infty } \end{array } \right]\ ] ] is nonsingular for all @xmath104 .
however , the limit point is clearly not a stationary point , thereby contradicting theorem [ pipa - conv ] .
in the preceding section , a small example has been presented for which pipa converges to a feasible but nonstationary limit point .
this example contradicts ( * ? ? ?
* theorem 6.1.17 ) , which establishes the convergence of pipa .
the example might be considered contrived in the sense that we choose a particular starting point and parameter values . however , for any different starting point it is possible to construct parameters such that pipa fails .
the reason for this apparent failure of pipa is the trust - region type constraint [ e - tr ] d_x ^2_2 c ( f^k + y^k^tw^k ) = : _ k in the direction - finding problem ( [ qp - dir ] ) .
the trust - region radius @xmath105 converges to zero as the iterates approach feasibility , thereby limiting the progress toward optimality in the controls @xmath14 .
this adverse behavior can be expected to occur whenever the iterates approach feasibility `` faster '' than optimality . linking the trust - region radius to the feasibility of
the problem is counterintuitive .
normally , @xmath105 is controlled by the algorithm , rather than by the iterates .
in particular , it should reflect how well the model problem ( [ qp - dir ] ) approximates the original problem , measured , for instance , by the agreement between actual reduction , @xmath106 and predicted reduction , @xmath107 table [ t - res ] shows that in this example , both the actual and the predicted reductions show near perfect agreement toward the end .
thus , one would expect @xmath105 to be increased rather than decreased .
the adverse situation in which ( [ e - tr ] ) limits progress toward optimality can be easily detected .
the lagrange multiplier of ( [ e - tr ] ) indicates whether progress can be made by relaxing this constraint . in the present example
, this multiplier converges to 1 .
one possible way to remedy the adverse behavior of pipa is to control the trust - region radius by using standard trust - region techniques .
the direction - finding problem then becomes [ qp - dir1 ] ll & f^k^t d + d_x^t q^k d_x + & x^k + d_x x + & f^k + f^k^t d = 0 + & y^k d_w + w^k d_y = - y^k w^k + e + & d_x _ k , where the trust - region is now controlled by the algorithm .
a typical updating scheme for @xmath105 is as follows .
first compute the ratio of actual over predicted reduction @xmath108 and then update @xmath105 according to @xmath109 & \mbox{if } \rho_k < \eta_1 \\ \left[\gamma_1 \delta_k,\delta_k\right ] &
\mbox{if } \rho_k \in [ \eta_1,\eta_2 ) \\ \left[\delta_k,\gamma_2 \delta_k\right ] & \mbox{if } \rho_k \geq \eta_2 \end{array } \right.\ ] ] in step 4 , where @xmath110 , and @xmath111 .
typical values for these constants are @xmath112 , @xmath113 , @xmath114 , and @xmath115 .
this results in a mixed trust - region / line - search algorithm .
convergence can be established under the assumption that the jacobian of @xmath2 , @xmath116 , \ ] ] remains nonsingular and has a bounded inverse along the lines of the coupled trust - region approach of dennis et al .
@xcite , since the qp subproblem can be regarded as a problem in @xmath13 only after eliminating @xmath117 using the linearization of @xmath2 .
i am grateful to roger fletcher and jorge nocedal for many fruitful discussions on mpccs and interior - point methods .
i am also grateful to jong - shi pang and an anonymous referee for useful comments on an earlier version of the paper .
this work was supported by the mathematical , information , and computational sciences division subprogram of the office of advanced scientific computing research , office of science , u.s .
department of energy , under contract w-31 - 109-eng-38 .
m. anitescu . on solving mathematical programs with complementarity constraints as nonlinear programs .
preprint anl / mcs - p864 - 1200 , mcs division , argonne national laboratory , argonne , il , 2000 .
submitted to siam j. optimization .
h. benson , d. f. shanno , and r. v. d. vanderbei .
interior - point methods for nonconvex nonlinear programming : complementarity constraints . technical report
orfe 02 - 02 , princeton university , operations research and financial engineering , 2002 .
s. p. dirkse , m. c. ferris , and a. meeraus .
mathematical programs with equilibrium constraints : automatic reformulation and solution via constraint optimization .
technical report na-02/11 , oxford university computing laboratory , july 2002 .
r. fletcher , s. leyffer , d. ralph , and s. scholtes . local convergence of sqp methods for mathematical programs with equilibrium constraints .
numerical analysis report na/209 , department of mathematics , university of dundee , dundee , uk , may 2002 .
j. huang and j .- s .
pang . a mathematical programming with equilibrium approach to the implied volatility surface of american options . technical report , dept . of mathematical sciences , the johns hopkins university , baltimore , maryland ,
december 1999 .
a. raghunathan and l. t. biegler .
barrier methods for mathematical programs with complementarity constraints ( mpccs ) . technical report , carnegie mellon university , department of chemical engineering , pittsburgh , december 2002 .
|
equilibrium equations in the form of complementarity conditions often appear as constraints in optimization problems .
problems of this type are commonly referred to as mathematical programs with complementarity constraints ( mpccs ) . a popular method for solving mpccs
is the penalty interior - point algorithm ( pipa ) .
this paper presents a small example for which pipa converges to a nonstationary point , providing a counterexample to the established theory .
the reasons for this adverse behavior are discussed .
* keywords : * nonlinear programming , interior - point methods , pipa , mpec , mpcc , equilibrium constraints .
* ams - msc2000 : * 90c30 , 90c33 , 90c51 , 49m37 , 65k10 .
|
quantum phase transitions ( qpts ) that happened at the zero temperature is purely a phenomenon of ground state ( gs ) transition driven by external parameters .
traditionally , qpts are described in terms of order parameter and symmetry breaking within the landau - ginzburg paradigm which have been extensively studied in condensed matter physics @xcite . in recent years
, qpt has also attracted a lot of attention in quantum - information science @xcite , in which one of the research focus is the role of quantum entanglement in characterizing qpts @xcite .
more recently , another concept in quantum information science , i.e. , the fidelity has been put forward to identify qpts from the perspective of the gs wave functions @xcite .
the gs fidelity is defined as the overlap between two ground states with only slightly different values of the external parameters @xcite and thus is a pure geometrical quantity . since no a priori knowledge of the order parameter is needed , the fidelity might be a potential universal criteria for characterizing the qpts @xcite .
an increasing interest has been drawn in the role of gs fidelity in detecting qpts for various many - body systems @xcite , since zanardi and paunkovic first exploited it to identify qpts in the xy spin chain @xcite where the fidelity shows a narrow drop at the transition point . remarkably , the success of fidelity analysis in dealing with the bose - hubbard model @xcite and spin systems @xcite implies that it may have practical relevance even for more complicate strongly interacting systems where no a simple description is possible .
the relation between the fidelity and berry phase @xcite has also been unveiled in terms of riemannian metric tensors @xcite .
generally one may expect that the gs fidelity shows an abrupt drop in the vicinity of the qpt point of the system as a consequence of the dramatic change of the structure of the gs .
this is true for the first - order qpts caused by a level crossing of gss due to the gs wavefunctions at the different sides of the level - crossing point are almost orthogonal ( orthogonal at the level - crossing point ) .
however , this is neither an obvious nor a general conclusion for a continuous qpt where no gs level crossing occurs and the gs evolves adiabatically " in the parameter space @xcite .
conventionally , qpts are characterized by singularities of the ground state energy : first - order qpts are characterized by discontinuities in the first derivative of the energy , whereas second - order ( @xmath0th - order ) qpts are characterized by discontinuities in the second ( @xmath0th ) derivative of the energy .
an important question is whether the singularity of gs energy is intrinsically related to the gs overlap ?
answering this question is no doubt significant for a deeper understanding of qpts , and the validity and limitation of fidelity as a measure of qpts .
so far the studies of fidelity as a measure of qpts are based on the analysis of particular many - body models @xcite .
the general connection between gs fidelity and qpts is not yet well established . in this work
, we shall discuss , in a general framework , how the gs fidelity can be related to a qpt characterized by non - analyticities of the gs energy .
our result shows that the singularity and scaling behaviors of the gs fidelity ( or it derivatives ) are directly related to its correspondences of the derivative of gs energy .
our paper is organized as follows . in section
ii , we reveal the general relation between the fidelity susceptibility and the 2nd order derivative of the gs energy .
the subsequent sections are devoted to two examples which exhibit the 2nd order and kosterlitz - thouless ( kt ) qpts , respectively .
a summary is given in the last section .
the general hamiltonian of a quantum many - body system undergoing qpts reads @xmath1 where @xmath2 is supposed to be the driving term with @xmath3 the control parameter . in terms of the eigenstates @xmath4 of @xmath5
, the hamiltonian can be reformulated as @xmath6 where @xmath7 is dimensions of the hilbert space .
the gs fidelity is defined as the overlap between @xmath8 and @xmath9 , i.e. @xmath10 where @xmath11 is the gs wavefunction corresponding to the parameter @xmath3 and @xmath12 is a small quantity @xcite .
it is obvious that the fidelity is dependent of @xmath12 .
the rate of change of fidelity is given by the second derivative of fidelity@xcite or fidelity susceptibility ( fs ) @xcite @xmath13 which is @xmath12 independent and sometimes a more effective quantity to detect the qpt .
first we discuss the first order qpt ( @xmath14qpt ) which is induced by gs level crossing .
we assume @xmath15 @xmath16 for @xmath17 with @xmath18 being the crossing point where the gs energy level @xmath19 crosses over the first excited level @xmath20 , therefore we must have @xmath21 for @xmath22 explicitly , for the @xmath14qpt , the gs energy is defined as @xmath23 the derivative of the gs energy is defined as @xmath24 in general , @xmath25 is not equal to @xmath26 therefore the first derivative of gs energy for a @xmath14qpt is not continuous at the transition point . accordingly , the gs fidelity at the transition point is defined as @xmath27 therefore we have @xmath28 which means that there appears a sharp drop in the transition point .
it is straightforward that the sudden drop in the transition point shares the same physical origin with the discontinuity of the first derivative of gs energy .
next we consider the continuous phase transition for which the gs of the hamiltonian is nondegenerate for a finite system .
for the case of @xmath29 order qpt , there is no level crossing for the gs energy , therefore we can always represent @xmath30 .
the first derivative of the gs energy is given by @xmath31 which is noting else but the hellmann - feynman theorem .
it is straightforward to get the second derivative of gs energy @xmath32 inserting the identity operator @xmath33 between @xmath2 and @xmath34 , we then get @xmath35 with @xmath36 we note that the term of @xmath37 is not included in the summation because it gives zero due to @xmath38 differentiating the eigenequation @xmath39 with respect to @xmath40 and taking the inner product with @xmath41 yields @xmath42 exploiting the hermiticity of @xmath43 to write @xmath44 it follows that for @xmath45 @xmath46 therefore , we have @xmath47 to see the relation of fs with the second derivative of gs energy , we rederive the expression of fs @xcite by expanding the wavefunction @xmath48 in the basis of eigenstates corresponding to the parameter @xmath3 , to the first order , which leads to @xmath49 where @xmath50 ^ 2}$ ] is the normalization constant .
it follows directly @xmath51 substituting it back into eq .
( [ fs ] ) , we get @xmath52 ^2}. \label{fsr}\ ] ] comparing eqs .
( [ fsr ] ) and ( [ 2de ] ) , we find that for both of them the singularities come from the vanishing energy gap in the thermodynamic limit .
however , a divergence does not necessarily occur in the critical point , for example , for the kt qpt as we shall discuss later , no singularity arises even the energy gap tends to zero in the thermodynamic limit as matrix elements @xmath53 may also vanish simultaneously . despite the apparent similarity between eqs .
( [ fsr ] ) and ( [ 2de ] ) , we note that fidelity susceptibility can not be expressed by the gs energy 2nd derivatives alone @xcite . nevertheless , the different power in the denominators in eqs .
( [ fsr ] ) and ( [ 2de ] ) shows that fidelity susceptibility might be a more sensitive tool to detect critical points because it can be more singular than the second derivative of gs energy .
therefore , for the case where the gs energy 2nd derivative is divergent at the critical point , it is no doubt that the fidelity susceptibility is divergent too .
however , for the case where the second derivative of gs energy is not divergent , ( for example the 3rd - order qpts and the kt phase transition ) , it is hard to judge whether the fidelity susceptibility becomes singular or not at these critical points just from the relation between eqs .
( [ fsr ] ) and ( [ 2de ] ) .
to understand the equivalence of the two approaches in characterizing the quantum criticality , we first apply them to study a specific example , say , the transverse field ising model with the hamiltonian given by @xmath54 where @xmath3 represents external magnetic field along the @xmath55 axis and the periodic boundary is assumed .
following the standard procedure @xcite , the model can be diagonalized by applying jordan - wigner transformation which maps the spin model to a fermion model and then using the bogoliubov transformation @xmath56 .
the result is given by @xmath57 with @xmath58 where the summation is over @xmath59 @xmath60 @xmath61 with the assumption of @xmath62 the ground state of @xmath43 is the vacuum @xmath63 of @xmath64 given by @xmath65 with the ground state energy @xmath66 , where @xmath67 and @xmath68 are , respectively , the vacuum and single excitation of the @xmath69th mode @xmath70 and @xmath71 is defined by @xmath72 it follows that the fidelity is given by @xmath73 with @xmath74 @xcite .
after some very simple algebras , one get @xmath75 in the thermodynamic limit with @xmath76 , the summation can be replaced by the integral and thus we get @xmath77 ^{3/2 } } d\varphi\ ] ] and @xmath78 ^2 } d\varphi\ ] ] where @xmath79 and @xmath80 .
both of the above integrals are divergent when @xmath81 .
their singular behavior can be analyzed in the vicinity of critical point @xmath82 with the asymptotic behavior described by @xmath83 and @xmath84 , respectively . as a function of @xmath3.,width=316 ]
to get further insight on how the two approaches work , we study both the scaling behaviors of fs and second derivative of gs energy . as shown in fig . 1
, we display the second derivatives of gs energy for different lattice sizes . despite no real divergence for finite @xmath85 ,
it is obvious that the curves exhibit marked anomalies with height of peak increasing with the lattice size .
the parameter @xmath86 labelling the position of peak approaches the critical point @xmath82 in a way of @xmath87 .
the value of peak diverges logarithmically with increasing lattice size as @xmath88 . by proper scaling scheme
, we can fit all the data of @xmath89 $ ] as a function of @xmath90 for different @xmath85 into a single curve as displayed in the inset of fig .
correspondingly , the fs exhibits similar scaling behaviors as shown in fig .
2 . around the critical point ,
all the data for different @xmath85 collapse into a single curve of @xmath91 $ ] as a function of @xmath90 . a related quantity has been used in @xcite to make scaling analysis for the quantum ising model .
it is clear that @xmath92 and @xmath93 exhibit similar critical behavior around the critical point .
having demonstrated the equivalent role of the two different approaches in identifying qpt on the transverse ising model , we now use this procedure to understand the more elusive kt - type qpt for which , contrary to the previous expectation , the gs fidelity and the fs turn out to be insensitive to qpt in some cases , for example , in the spin @xmath95 model @xcite .
we now consider the @xmath94 model which provides a further test for the intrinsic relationship between the derivatives of gs energy and fs .
the model is defined by @xmath96 with @xmath3 the exchange anisotropy parameter and it is related to @xmath97 by a unitary transformation .
it is well known that the model is in a critical phase for @xmath98 , an antiferromagnetic phase at @xmath99 , and ferromagnetic phase at @xmath100 the qpt at the ferromagnetic isotropic point @xmath81 is a 1qpt caused by the level crossing of gs .
here we shall focus on the qpt at the antiferromagnetic isotropic point @xmath101 which is known to be of kt type . in the whole parameter regime
the model is integrable .
the eigenstate of @xmath5 has the form of @xmath102 with the coefficients having the bethe - ansatz form @xcite @xmath103\ ] ] where @xmath104 is any permutation of @xmath105 and the phase @xmath106}{\cos [ ( p+q)/2]-\lambda \cos [ ( p - q)/2]}\right ) $ ] .
the eigenenergies @xmath107 are determined by a set of quasi - momentum @xmath108 ( @xmath109 for @xmath61 down spins which are the solution of bethe ansatz equations ( baes ) @xmath110 , where @xmath111 are integer or half - odd integer .
model as a function of @xmath112 .
the data labelled by @xmath113 and @xmath114 are obtained by ed and they comply with the data by bethe - ansatz method exactly.,width=345 ] by numerically solving the baes , we can get the gs state energy for different sizes of @xmath85 . as shown in fig.3
, we calculate the @xmath115nd derivative of gs energy by both bethe - ansatz and exact diagonaliztion ( ed ) methods and the fs by ed .
our numerical results show that there is no any singularity around the critical point @xmath116 for both the fs and @xmath115nd order derivative of gs energy for different sizes . for the gs energy derivatives ,
this is true even in the thermodynamic limit where actually , as proven in @xcite , all the @xmath0th order derivatives of gs energy are continuous .
thus , in view of its connection to the derivatives of gs energy , we can understand why the gs fidelity failed to reproduce quantum critical behavior by the desired singularities in the critical point for the kt transition .
it is also natural to explain why the finite size scaling of fs has a clear distinction between noncritical and critical phases because the gs energies are known to fulfil quite different finite size scaling in critical or noncritical phase @xcite .
the gs energy in the critical regime fulfills the finite - size correction @xmath117 with @xmath118 being the conformal anomaly number and @xmath119 the `` sound velocity '' for the model @xcite . here
@xmath120 is defined by @xmath121 . on the other hand ,
the conformal invariance is broken for @xmath122 and the gs energy fulfills a different finite size scaling @xcite .
bearing in mind that the equivalent role of fs and derivative of gs energy , we expect that the fs exhibits similar scaling behavior in the critical regime as its counterpart of derivative of gs energy does and there is a deviation from the scaling behavior when @xmath122 .
indeed , the fs has been shown to scale differently in critical and noncritical phases based on the results by ed method @xcite .
this feature can be used as an indicator of kt qpt occurring in this model . before ending the discussion
, we would like to give a remark to the problem whether the fidelity approach is suitable to detect the kt transition ? as no rigorous result of the fs even for the exactly solvable model is available in the thermodynamic limit , therefore this question is still in a doubtful status @xcite and remains an open problem .
we have shown the intrinsic relationship of the fidelity and derivatives of gs energy and revealed their equivalent role in identifying the qpts . within our framework
, the singularities and scaling behavior of fidelity for various systems can be well understood on general grounds . through concrete examples
, we display that divergence of the fs or derivatives of gs energy can be applied to identify the 2nd order qpt , whereas for the kt transition the criticality is not a sufficient condition to ensure divergence of the fs or derivatives of gs energy .
the authors are grateful to m. f. yang for useful discussions by correspondence and the referee for bringing our attention to ref .
this work is supported by nsf of china under grant no .
10574150 , most grant 2006cb921300 and programs of cas .
p. zanardi , m. cozzini , and p. giorda , j. stat .
* 2 * , l02002 ( 2007 ) ; m. cozzini , p. giorda , and p. zanardi , phys . rev .
b , * 75 * , 014439 ( 2007 ) ; m. cozzini , r. ionicioiu , and p. zanardi , _ ibid . _ * 76 * , 104420 ( 2007 ) .
l. c. venuti and p. zanardi , phys .
lett . * 99 * , 095701 ( 2007 ) .
p. zanardi , p. giorda , and m. cozzini , phys .
lett . * 99 * , 100603 ( 2007 ) . after the submision of the present paper ( arxiv : 0801.0020v1 )
, there appears a paper ( l. campos venuti , m. cozzini , p. buonsante , f. massel , n. bray - ali , and p. zanardi , arxiv : 0801.2473v1 ) , where an expression of fidelity susceptibility including the third order contributions is given .
|
the notion of fidelity in quantum information science has been recently applied to analyze quantum phase transitions from the viewpoint of the ground - state ( gs ) overlap for various many - body systems . in this work ,
we unveil the intrinsic relation between the gs fidelity and the derivatives of gs energy and find that they play equivalent role in identifying the quantum phase transition .
the general connection between the two approaches enables us to understand the different singularity and scaling behaviors of fidelity exhibited in various systems on general grounds .
our general conclusions are illustrated via several quantum spin models which exhibit different kinds of qpts .
|
( also hd 331102 , bd+30 3704 , tyc 2656 - 3703 - 1 , @xmath11 ) is an early type ( b9.5+b9.5 ) detached , eccentric , double - lined eclipsing binary with a relatively long period of 15.33 days @xcite and nearly identical components . the first photoelectric light curve was obtained and analyzed by @xcite , but the masses could not be determined dynamically because no spectroscopic observations were available at the time .
this was remedied by @xcite , who reported radial - velocity measurements for both components and analyzed them in conjunction with the @xmath12-band light curve from @xcite to obtain the absolute masses ( @xmath132.2 @xmath14 ) and radii ( @xmath131.8 @xmath15 ) with relative errors of about 4% and 2% , respectively .
the system is noteworthy in that it presents a periastron advance that is dominated by the general relativistic effect , estimated to be several times larger than the classical effects due to tidal and rotational distortions .
however , there has been some disagreement over the precise rate of apsidal motion , which is fairly slow and difficult to determine , and how closely it conforms to theoretical expectations .
some authors have obtained good agreement with the predicted motion , while others have measured a rate of precession that is too slow , and argued that may belong to a small group of binaries including di her and as cam that display similar discrepancies .
past speculations about possible shortcomings of general relativity or alternative theories of gravitation that might explain the puzzle ( e.g. , * ? ? ?
* ) have largely fallen out of favor , and at least in the case of di her the `` anomalous '' apsidal motion has now been shown to be caused by the tilted spin axes of the stars relative to the axis of the orbit @xcite .
the motivation for this paper is twofold .
firstly , we note that the precision of the mass and radius estimates by @xcite was limited by the quality and quantity of his spectroscopic observations . in particular
, the masses are not quite precise enough for a meaningful comparison with modern stellar evolution models for this important binary system ( see , e.g. , * ? ? ?
* ) , a test that was not performed in the original study by that author .
to this end we have obtained a much richer set of high - resolution spectra that allows substantial improvement in the absolute dimensions of the two stars .
we also bring to bear additional light curves obtained more recently to supplement the only existing set of photometric measurements by @xcite .
secondly , we wish to revisit the determination of the apsidal motion and the comparison with internal structure theory by taking advantage of all existing measurements ( light curves , radial - velocity measurements , eclipse timings ) in a self - consistent way .
previously this has been done using only the measured times of eclipse .
the paper is organized as follows .
section [ sec : spectroscopy ] and section [ sec : photometry ] describe our new spectroscopic and photometric observations of , and the times of minimum light that have the longest time coverage are presented in section [ sec : timings ] .
our combined analysis of all the data is described in section [ sec : analysis ] , followed by a summary of the inferred physical properties of the components in section [ sec : dimensions ] .
then in section [ sec : evolution ] we discuss the comparison of the mass , radius , and temperature determinations for against current models of stellar evolution , and our new measurement of the rate of apsidal motion of the binary is presented in section [ sec : apsidal ] .
we end with a discussion of these results and final remarks in section [ sec : discussion ] .
was observed spectroscopically at the harvard - smithsonian center for astrophysics ( cfa ) with the digital speedometer ( ds ; * ? ? ?
* ) on the 1.5 m tillinghast reflector at the fred l. whipple observatory on mount hopkins ( az ) .
this instrument ( now decommissioned ) was an echelle spectrograph with a resolving power of @xmath16 equipped with a photon - counting intensified reticon detector limiting the output to a single order 45 wide centered on the b triplet at 5187 .
we gathered 72 exposures between 2000 july and 2004 october with signal - to - noise ratios ranging from 20 to 49 per resolution element of 8.5 .
one of the observations was obtained during an eclipse and was excluded from further analysis .
reductions were performed with a custom pipeline , and the wavelength calibration was based on exposures of a thorium - argon lamp before and after each science exposure .
exposures of the dusk and dawn sky were taken regularly for the purpose of monitoring instrumental drifts .
all spectra are clearly double - lined .
radial velocities ( rvs ) were determined by cross - correlation using the two - dimensional algorithm todcor @xcite .
templates ( one for each component ) were taken from a large library of synthetic spectra based on model atmospheres by r. l. kurucz ( see * ? ? ?
* ; * ? ? ?
* ) , computed for a range of temperatures ( @xmath17 ) , surface gravities ( @xmath18 ) , rotational broadenings ( @xmath19 when seen in projection ) , and metallicities ( [ m / h ] ) .
the optimum templates were selected by cross - correlating each of our 71 spectra against synthetic spectra covering a wide range of parameters , and seeking the best match as measured by the maximum cross - correlation coefficient averaged over all exposures and weighted by the strength of each spectrum ( see * ? ? ? * ) .
experience has shown that the narrow wavelength coverage of our spectra introduces strong correlations between @xmath17 , @xmath20 , and [ m / h ] such that optimal matches of similar quality can be obtained by slightly increasing or decreasing these three parameters in tandem .
we therefore assumed initially that the metallicity is solar , and held @xmath18 fixed at values near those determined later in our analysis ( section [ sec : dimensions ] ) . because the grid spacing of our library in @xmath18 is 0.5 dex and the actual values ( @xmath214.25 for both stars ) are intermediate between two grid points , we repeated the determinations using @xmath18 values of 4.0 and 4.5 , and interpolated the results .
similarly , the comparison with stellar evolution models described in section [ sec : evolution ] points to a composition intermediate between @xmath22 } = 0.0 $ ] and @xmath23 , so we repeated the determinations at the lower metallicity ( for each of the two values of @xmath18 ) and again interpolated .
the resulting best - fit temperatures of the slightly larger and more massive primary component ( star 1 ) and of the secondary ( star 2 ) are 10650 k and 10350 k , respectively , with estimated uncertainties of 200 k. the @xmath19 values are @xmath24 for both stars .
templates near these best - fit values were used to derive the radial velocities .
the flux ratio between the secondary and primary as determined with todcor is @xmath25 at the mean wavelength of our observations , 5187 .
systematic errors in the radial velocities that may result from spectral lines shifting in and out of the narrow spectral window as a function of orbital phase were investigated by means of numerical simulations as described by @xcite .
briefly , we generated artificial composite spectra matching each of the real spectra by combining the two templates scaled by the above flux ratio and with the proper relative doppler shifts as determined from a preliminary spectroscopic orbital solution .
we then processed these artificial spectra with todcor in the same way as the real spectra , and compared the input and output doppler shifts .
the differences were applied as corrections to the raw velocities measured from the real spectra , and were typically smaller than 0.5 , which is only about 1/3 of our internal errors .
the final radial velocities in the heliocentric frame are listed in table [ tab : rvs ] and include these corrections for systematics as well as small run - to - run adjustments to remove instrumental drifts , as described above .
typical uncertainties are about 1.5 .
lccccccc 51740.7766 & @xmath2657.26 & 1.35 & @xmath260.27 & 27.09 & 1.30 & @xmath260.59 & 0.1654 + 51743.8481 & @xmath2696.49 & 2.72 & @xmath272.75 & 69.96 & 2.62 & @xmath261.32 & 0.3657 + 51743.8753 & @xmath2698.75 & 1.62 & @xmath270.34 & 69.06 & 1.56 & @xmath262.07 & 0.3675 + 51800.7427 & @xmath2637.09 & 1.31 & @xmath260.95 & 4.91 & 1.26 & @xmath261.26 & 0.0751 + 51802.6632 & @xmath2664.62 & 1.39 & @xmath271.02 & 36.69 & 1.34 & @xmath270.09 & 0.2003 in addition to our own observations we have made use in our analysis below of the 16 radial - velocity measurements by @xcite , which , although less numerous and of lower precision than our own , were obtained much earlier ( 19821994 ) and can help to constrain the periastron advance .
the @xmath12-band observations of @xcite , which we reanalyze here , consist of 531 differential measurements obtained between 1981 may and 1983 july with a 0.5 m reflector at the crimea observatory of the sternberg astronomical institute .
the comparison star was bd+30 3702 ( hd 331103 ) and the check star was bd+31 3728 ( hd 331101 ) .
the reported internal precision of these measurements is 0.009 mag .
more recent observations of were obtained between 2007 may and
2014 november in the course of the kilodegree extremely little telescope transiting planet program ( kelt ; * ? ? ?
kelt is an all - sky photometric survey to discover transiting planets around bright host stars ( @xmath28 ) .
it uses two telescopes : kelt - north located in sonoita ( az ) , and kelt - south at the south african astronomical observatory ( saao ) .
both telescope systems are based on a mamiya 645-series wide - angle 42 mm lens with an 80 mm focal length ( @xmath29/1.9 ) giving a field of view of @xmath30 .
the observations of reported here were gathered at the northern site with a @xmath31 pixel apogee ap16e ccd camera ( 9 @xmath32 m pixels , corresponding to 23 on the sky ) and a 1020 minute cadence .
is located in kelt - north field 11 ( @xmath33 = 19@xmath3427@xmath3500@xmath36 , @xmath37 = @xmath3839@xmath3956 , j2000 ) .
the telescopes are mounted on a paramount me german equatorial mount causing images taken east of the meridian to have a 180 rotation compared to the images to the west @xcite . as the optics for each telescope
are not perfectly axisymmetric , the point spread function ( psf ) of the same star in the corner of the images is not the same in each orientation . in addition , the psf full - width - at - half - maximum ( fwhm ) varies between 3 and 6 pixels depending on where the star is positioned on the ccd .
therefore , the east and west images have been treated as independent time series throughout the reduction and detrending process .
because of the large psf ( @xmath2112 on the sky ) , light from neighboring stars is likely to affect the photometry and must be taken to account , as described in section [ sec : analysis ] .
the passband of these observations resembles that of a very broad @xmath40-band filter
. reductions and detrending were carried out with a dedicated pipeline that produces three types of light curves for each target star in the program : a raw extracted light curve , a `` scaled '' version that is the raw light curve with a 90-day median smoothing applied , and a detrended version that uses the trend filtering algorithm ( tfa ; * ? ? ?
for a detailed description of the kelt data acquisition , reduction , and post - processing , see @xcite and @xcite . because the tfa procedures are tuned for the detection of very shallow , transit - like signals characteristic of exoplanets ,
the results for objects with very deep ( @xmath210.7 mag ) eclipses such as those in are less than optimal .
therefore , for the analysis in this paper we have made use of the `` scaled '' light curves .
the time series east of the meridian consists of 3885 measurements with a typical internal photometric precision of 0.017 mag , and the one to the west has 2940 measurements with typical uncertainties of 0.013 mag .
these observations are reported in table [ tab : kelteast ] and table [ tab : keltwest ] , respectively .
lcc 54257.770944 & 14.941 & 0.016 + 54257.775454 & 14.926 & 0.015 + 54257.779964 & 14.903 & 0.015 + 54257.784484 & 14.934 & 0.015 + 54257.788995 & 14.915 & 0.016 lcc 54250.936923 & 14.615 & 0.013 + 54250.942503 & 14.637 & 0.013 + 54276.867035 & 14.658 & 0.011 + 54362.631612 & 14.637 & 0.011 + 54363.630845 & 14.665 & 0.012
numerous times of minimum light have been recorded for and used over the years to improve the ephemeris .
the more recent ccd measurements carry essentially all of the weight , and older photographic or visual timings that have occasionally been included have such large scatter that they are of little use for determining the apsidal motion .
prior to the publication of the photoelectric light curve of @xcite the most complete light curve of was the one reported by @xcite , based on his extensive series of 161 photographic plates collected with a 14 cm astrograph at ostrowik near warsaw ( poland ) between 1955 and 1959 . from these observations
karpowicz reported three average times of minimum light for the primary and three for the secondary , each determined from plates restricted to time spans of no more than two years .
though perhaps better than other photographic estimates , even those timings give very large residuals by modern standards .
however , it is still possible to extract useful information from this material in the form of one average time of primary minimum and one average time of secondary minimum for the entire series .
this can be done by using all of karpowicz original photographic measurements together , which allows more complete coverage of the eclipses and therefore a better definition of the minima .
the apsidal motion is slow enough that any change in the separation between primary and secondary eclipse over the four - year interval of these observations is small compared to the uncertainties .
we proceeded as follows .
we phase - folded the 161 brightness measurements with a preliminary ephemeris based on the analysis below , and then fit the primary and secondary eclipses separately using as a template a model light curve from a preliminary solution to the @xmath12-band data of @xcite . after applying a vertical offset to the template by eye , we allowed only a horizontal ( phase ) shift to fit each eclipse ( keeping the shape fixed ) , and finally converted the phase shifts back to time units .
we assigned these measurements to a reference epoch closest to the average time of all photographic observations , obtaining @xmath41 ( hjd ) and @xmath42 ( hjd ) .
table [ tab : timings ] lists these two timings along with all other published photoelectric / ccd timings .
we note that , although the table also lists the two timings reported by @xcite , we do not use them here because our analysis incorporates the full @xmath12-band light curve from this author , which contains the same information .
ll@c@cc@c@ 36262.3278 & 0.0031 & 1 & @xmath260.00007 & 1958.1583 & 1 + 36269.3158 & 0.0031 & 2 & @xmath270.00151 & 1958.1775 & 1 + 44882.2148 & 0.0007 & 1 & @xmath260.00026 & 1981.7583 & 2 + 44889.2196 & 0.0005 & 2 & @xmath260.00006 & 1981.7775 & 2 + 46998.8424 & 0.0010 & 1 & @xmath270.00024 & 1987.5533 & 3 + 48616.3400 & & 2 & @xmath270.00823 & 1991.9818 & 4 + 48839.3870 & 0.003 & 1 & @xmath260.00046 & 1992.5924 & 5 + 49168.4947 & 0.0007 & 2 & @xmath260.00183 & 1993.4935 & 6 + 49168.4955 & 0.0009 & 2 & @xmath260.00103 & 1993.4935 & 6 + 49560.2668 & 0.0008 & 1 & @xmath260.00091 & 1994.5661 & 7 + 49889.3770 & 0.001 & 2 & @xmath260.00130 & 1995.4672 & 8 + 49904.7145 & 0.00012 & 2 & @xmath260.00171 & 1995.5091 & 9 + 49935.3911 & 0.0006 & 2 & @xmath260.00093 & 1995.5931 & 10 + 49935.3974 & 0.0012 & 2 & @xmath270.00537 & 1995.5931 & 11 + 50296.4906 & 0.0018 & 1 & @xmath270.00477 & 1996.5818 & 12 + 51070.3967 & & 2 & @xmath260.00066 & 1998.7006 & 13 + 51109.3918 & & 1 & @xmath260.00154 & 1998.8074 & 13 + 51385.4740 & 0.0007 & 1 & @xmath260.00113 & 1999.5632 & 14 + 52926.28338 & 0.0001 & 2 & @xmath260.00108 & 2003.7817 & 15 + 53524.46196 & 0.0001 & 2 & @xmath260.00099 & 2005.4195 & 15 + 53578.7911 & 0.0001 & 1 & @xmath260.00052 & 2005.5682 & 16 + 53846.55486 & 0.0007 & 2 & @xmath260.00420 & 2006.3013 & 15 + 54659.46742 & 0.0002 & 2 & @xmath260.00086 & 2008.5269 & 17 + 54659.46762 & 0.0002 & 2 & @xmath260.00066 & 2008.5269 & 17 + 54659.46809 & 0.0001 & 2 & @xmath260.00019 & 2008.5269 & 15 + 54782.1712 & 0.0003 & 2 & @xmath260.00036 & 2008.8629 & 15 + 55066.5740 & 0.0024 & 1 & @xmath270.00827 & 2009.6415 & 18 + 55741.42898 & 0.0007 & 1 & @xmath260.00336 & 2011.4892 & 19 + 55741.4299 & 0.0064 & 1 & @xmath260.00244 & 2011.4892 & 20 + 55741.43011 & 0.0010 & 1 & @xmath260.00223 & 2011.4892 & 19 + 55794.4734 & 0.0017 & 2 & @xmath260.00022 & 2011.6344 & 21 + 56876.4377 & 0.0026 & 1 & @xmath270.00242 & 2014.5967 & 22 + 56929.4809 & 0.0020 & 2 & @xmath270.00195 & 2014.7419 & 23 + 57198.5271 & 0.0026 & 1 & @xmath260.00361 & 2015.4785 & 24 in general
one expects the times of secondary minimum to be more precisely determined than those of primary minimum for , on account of the much shorter duration of the secondary eclipse ( 6 hours ) compared to the primary ( 16 hours ) caused by the large eccentricity and orientation of the orbit .
however , this does not seem to be reflected in the formal uncertainties shown in table [ tab : timings ] , possibly due to the heterogeneous nature of these measurements .
many studies have shown that formal timing errors tend to be underestimated , and our analysis below supports this .
the various data sets available for are complementary in many ways .
for example , the light curve of @xcite is of high quality but has incomplete coverage of the primary eclipse . the kelt data , on the other hand , have somewhat lower precision but offer essentially complete coverage of all phases , aiding in determining some of the geometric properties of the orbit .
these two photometric time series provide strong constraints on the location of the line of apsides at two epochs some three decades apart .
the times of minimum light from the previous section provide additional constraints on the apsidal motion over a longer time span of 57 years ( see figure [ fig : history ] ) .
finally the radial velocities yield constraints of a different nature at other times , particularly on the eccentricity of the orbit , so that the optimum approach is to combine all of these data in a global fit to solve for all orbital parameters simultaneously , including the apsidal motion . for the analysis of the light curves of this well - detached system we have adopted the nelson - davis - etzel model @xcite , implemented originally in the widely used ebop code .
the particular version of the light - curve generator in ebop used here is a rewrite due to @xcite that also handles apsidal motion , and is especially useful within the framework of the markov chain monte carlo methodology we apply here .
the parameters solved for include the sidereal period @xmath43 , a reference time of primary eclipse @xmath44 chosen to be near the mean of all our times of observation to minimize correlations , the sum of the relative radii @xmath45 normalized to the semimajor axis , the radius ratio @xmath46 , the cosine of the inclination angle @xmath47 , the eccentricity parameters @xmath48 and @xmath49 at @xmath44 , the rate of apsidal motion @xmath50 , the central surface brightness ratio @xmath51 , the fraction of third light divided by the total light ( @xmath52 ) , and a magnitude zero point @xmath53 .
light travel time across the system was accounted for , although the effect is hardly noticeable . for the limb darkening effect we adopted the linear law with a coefficient @xmath54 ( assumed to be the same for the two nearly identical components ) , as experiments with higher - order prescriptions indicated no improvement .
the gravity darkening input required by the code is the product of two values : the exponent @xmath55 of the law relating the emergent bolometric flux to the local surface gravity , and a separate wavelength - specific gravity darkening coefficient , @xmath56 . as affected by local gravity
@xmath57 using either the blackbody formula or model atmospheres , and therefore expect the input values for gravity darkening to be the exponent ( @xmath55 in recent usage ) of the bolometric law @xmath58 for each star @xcite , ebop has no knowledge of the passband of the observations , or in fact of the absolute temperatures of the stars ( by design ) .
therefore , it is up to the user to supply values appropriate for the wavelength @xmath59 and stellar temperatures , as is the case for the limb - darkening coefficients .
ebop adopts a simple taylor expansion of the wavelength - dependent flux as a function of local gravity of the form @xmath60 $ ] ( see , e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , retaining only the linear term consistent with the level of other approximations and the intended application of the model to simple , well - detached binary systems .
the input required by ebop is the _ product _ of @xmath55 and the gravity darkening coefficient @xmath56 , which depends on wavelength and also temperature .
expressions to compute @xmath56 are provided in the above references using a blackbody approximation , which is sufficient for our purposes given that the gravity darkening effect is essentially negligible for nearly spherical stars such as those in .
note that @xmath61 is unity for the total ( bolometric ) flux , but varies significantly as a function of @xmath17 and @xmath59 and can be larger or smaller than unity .
although the prescription for gravity darkening in terms of the coefficient @xmath61 is described in the original documentation for ebop @xcite , where the implicit assumption is also made that @xmath55 is always unity , that document is unpublished and has generally been difficult to obtain , which may have contributed to the confusion . ] here we assume @xmath62 , as appropriate for radiative stars , and the coefficients @xmath56 were set for each passband to values appropriate for the stellar temperatures : 0.67 and 0.68 for the primary and secondary in @xmath12 , and 0.59 and 0.61 for the kelt passband .
separate values of the wavelength - dependent parameters @xmath63 and @xmath54 were considered for the @xcite data and for the two kelt light curves , and magnitude zero points were solved independently for each of the three light curves .
inclusion of the third light parameter was motivated in part by the report of the presence of a close visual companion to in the _ tycho _ double star catalogue @xcite , at an angular separation of 075 in position angle 62 and with a reported brightness difference of about 1.2 mag in the @xmath64 passband .
we note , however , that subsequent observations by @xcite using speckle interferometry did not detect this companion , even though they were sensitive to it .
one value of @xmath52 was included for the @xmath12-band data , and two additional values were used for the kelt data sets because their much larger psf ( @xmath2112 ; section [ sec : photometry ] ) is different for the east and west ( see section [ sec : photometry ] ) . for the radial velocities we assumed pure keplerian orbital motion , and solved for the velocity semi - amplitudes @xmath65 and @xmath66 , the center - of - mass velocity @xmath67 , as well as for a possible offset @xmath68 between the velocity zero points for the primary and secondary .
the latter may arise , for example , from template mismatch in the cross - correlation procedure .
the parameters @xmath67 and @xmath68 were solved independently for our own rvs and those of @xcite .
the times of minimum light were incorporated using the formalism of @xcite for the ephemeris curve solution .
we note that the ebop light - curve generator used here assumes @xmath44 is a time of inferior conjunction , rather than a time of minimum light , but the difference for this system is negligible .
our method of solution used the emcee code of @xcite , which is a python implementation of the affine - invariant markov chain monte carlo ensemble sampler proposed by @xcite .
uniform priors over suitable ranges were used for most parameters , and modified jeffreys priors were used for third light and the jitter parameters @xcite , although the results are insensitive to these assumptions .
relative weighting between the different data sets was handled by including additional adjustable parameters to inflate the observational errors , which were solved for self - consistently and simultaneously with the other parameters ( see * ? ? ?
* ) . to inflate the uncertainties of the three light curves we adopted simple scale factors @xmath69 , @xmath70 , and @xmath71 , as is customary , and for the radial velocities and the times of minimum we used separate `` jitter '' terms for the primary and secondary ( @xmath72 , @xmath73 , @xmath74 , @xmath75 ) that we added quadratically to the published uncertainties . for the rvs of @xcite , which have no published uncertainties , we assumed errors of 2.5 .
the rv jitter terms were independent for the cfa velocities and those of @xcite .
initial solutions converged to a value of the radius ratio @xmath76 larger than unity , which for main - sequence stars such as those in is inconsistent with a derived mass ratio that is smaller than unity .
this is a common problem in fitting light curves of similar stars with partial eclipses , and is due to strong correlations among several of the parameters . in this case
, external information must be used to lift the degeneracies , such as a light ratio from spectroscopy ( see , e.g. , * ? ? ?
* ) , which is very sensitive to the radius ratio : @xmath77 .
we therefore imposed a gaussian prior on the @xmath12-band light ratio generated at each step in our monte carlo solution , set by the spectroscopic determination from section [ sec : spectroscopy ] , @xmath25 . while this measurement is not strictly in the @xmath12 passband , the difference is negligible for our purposes because of the similarity in temperature between the components .
lc @xmath43 ( days ) & @xmath78 + [ + 1ex ] @xmath44 ( hjd-2,400,000 ) & @xmath79 + [ + 1ex ] @xmath80 & @xmath81 + [ + 1ex ] @xmath82 & @xmath83 + [ + 1ex ] @xmath45 & @xmath84 + [ + 1ex ] @xmath46 & @xmath85 + [ + 1ex ] @xmath47 & @xmath86 + [ + 1ex ] @xmath87 & @xmath88 + [ + 1ex ] @xmath89 & @xmath90 + [ + 1ex ] @xmath50 ( deg century@xmath10 ) & @xmath91 + [ + 1ex ] @xmath92 & @xmath93 + [ + 1ex ] @xmath94 & @xmath95 + [ + 1ex ] @xmath96 & @xmath97 + [ + 1ex ] @xmath98 & @xmath99 + [ + 1ex ]
@xmath100 & @xmath101 + [ + 1ex ] @xmath102 & @xmath103 + [ + 1ex ] @xmath104 & @xmath105 + [ + 1ex ] @xmath106 & @xmath107 + [ + 1ex ] @xmath69 & @xmath108 + [ + 1ex ] @xmath70 & @xmath109 + [ + 1ex ] @xmath110 & @xmath111 + [ + 1ex ] @xmath74 ( days ) & @xmath112 + [ + 1ex ] @xmath75 ( days ) & @xmath113 + [ + 1ex ] @xmath114 ( ) & @xmath115 + [ + 1ex ] @xmath116 ( ) & @xmath117 + [ + 1ex ] @xmath118 ( ) & @xmath119 + [ + 1ex ] @xmath120 ( ) & @xmath121 + [ + 1ex ] @xmath65 ( ) & @xmath122 + [ + 1ex ] @xmath66 ( ) & @xmath123 + [ + 1ex ] @xmath72(cfa ) ( ) & @xmath124 + [ + 1ex ] @xmath73(cfa ) ( ) & @xmath125 + [ + 1ex ] @xmath72(lacy ) ( ) & @xmath126 + [ + 1ex ] @xmath73(lacy ) ( ) & @xmath127 + + @xmath128 ( days ) & @xmath129 + [ + 1ex ] @xmath130 & @xmath131 + [ + 1ex ] @xmath132 ( deg ) & @xmath133 + [ + 1ex ] @xmath50 ( @xmath134 rad day@xmath10 ) & @xmath135 + [ + 1ex ] @xmath136 ( years ) & @xmath137 + [ + 1ex ] @xmath138 ( deg ) & @xmath139 + [ + 1ex ] @xmath140 & @xmath141 + [ + 1ex ] @xmath142 & @xmath143 + [ + 1ex ] @xmath144 & @xmath145 + [ + 1ex ] @xmath146 ( @xmath15 ) & @xmath147 + [ + 1ex ] @xmath148 & @xmath149 + [ + 1ex ] @xmath150 & @xmath151 + [ + 1ex ] phase of min ii at @xmath44 & @xmath152 the results of our global fit are presented in table [ tab : results ] , where we list for each parameter the mode of the posterior distributions along with the 68.3% credible intervals .
additional properties derived from the fitted quantities were computed by combining the corresponding markov chains , link by link , and are reported as well .
as anticipated , third light is highly significant for both of the kelt light curves , indicating an appreciable amount of flux is coming from nearby field stars .
contamination is larger to the east ( @xmath2118% ) compared to the west ( @xmath2112% ) , which is also as expected .
the fitted @xmath52 for the @xmath12-band light curve of @xcite is formally below 0.2% and not significantly different from zero .
therefore , we find no evidence for the @xmath153 mag companion reported in the _ tycho _ double star catalogue , which if real should contribute about 25% of the light in @xmath12 . this star would be expected to be even more prominent in the redder kelt passband , and yet the corresponding @xmath52 values are smaller than 25% .
the radial - velocity offsets @xmath68 between the primary and secondary for both rv data sets are also not statistically significant .
repeating the fit with @xmath98 and the velocity offsets set to zero leads to nearly identical values for all elements .
nevertheless , we have chosen here to retain the results that include these extra parameters , to be conservative , so that any uncertainty in their determination is propagated through to the rest of the adjusted quantities .
the linear limb - darkening coefficient for the @xmath12 band , @xmath105 , is somewhat lower than the value of @xmath210.45 expected from theory for the mean temperature of the components ( e.g. , * ? ? ?
the coefficient obtained for the kelt light curves , @xmath107 , is close to the value of 0.38 predicted for the @xmath40 band but this is likely accidental as the passbands are not exactly the same . a graphical representation of the photometric observations and the fitted light curves are given in figure [ fig : khaliullin ] for the @xmath12-band data of @xcite , and in figure [ fig : kelteast ] and figure [ fig : keltwest ] for the kelt measurements . in each case the residuals are shown below each panel .
the resulting scatter of the photometric measurements is 0.015 mag in @xmath12 , 0.025 mag for kelt east , and 0.017 mag for kelt west .
the cfa rv observations along with those of @xcite are shown with the fitted model in figure [ fig : rvs ] .
for the cfa data ( 71 measurements ) the rms residuals from the fit are 1.51 and 1.63 for the primary and secondary , respectively .
the 16 measurements by @xcite show a scatter of 3.54 and 4.10 .
the absolute masses and radii of are listed in table [ tab : dimensions ] .
the relative uncertainties are smaller than 1% in both properties and represent a significant improvement over those obtained by @xcite .
our temperature estimates from section [ sec : spectroscopy ] are marginally different for the two components , and are somewhat higher than those obtained by @xcite , @xmath154 k , and by @xcite , @xmath155 k , both of whom assumed identical temperatures for the two stars .
strmgren photometry obtained by @xcite along with the calibration by @xcite for b - type stars yields a mean reddening of @xmath156 , from which @xmath157 for the combined light .
tabulations by @xcite and @xcite then yield mean temperatures of 10670 k and 10920 k , respectively .
alternatively , the corresponding reddening in the johnson system , @xmath158 , together with the measured @xmath159 index of @xcite give @xmath160 , from which we obtain mean temperatures of 10250 k @xcite , 10890 k @xcite , and 10460 k @xcite .
the average of the above five photometric estimates , @xmath161 k , is consistent with our spectroscopic determinations ( 10650 k for the primary and 10350 k for the secondary ) , supporting their accuracy .
no spectroscopic metallicity is available for ; our own spectra are unsuitable for a detailed analysis because of the narrow wavelength range , the relatively low signal - to - noise ratios , and the degeneracy with temperature mentioned earlier .
table [ tab : dimensions ] lists our measured projected rotational velocities for the components , which are considerably smaller than earlier estimates of @xmath162 by @xcite and @xmath163 by @xcite .
also listed are the expected rotational velocities assuming either pseudo - synchronous rotation @xcite or that the stars are synchronized to the mean orbital motion .
the measurements agree well with the pseudo - synchronous values .
lcc mass ( @xmath14 ) & @xmath164 & @xmath165 + [ + 1ex ] radius ( @xmath15 ) & @xmath166 & @xmath167 + [ + 1ex ] @xmath18 ( cgs ) & @xmath168 & @xmath169 + [ + 0.5ex ] temperature ( k ) & 10650 @xmath170 200 & 10350 @xmath170 200 + [ + 0.5ex ] @xmath171 & @xmath172 & @xmath173 + [ + 0.5ex ] @xmath174 ( mag ) & @xmath260.39 @xmath170 0.11 & @xmath260.32 @xmath170 0.11 + [ + 0.5ex ] @xmath175 ( mag ) & @xmath176 & @xmath177 + [ + 0.5ex ] @xmath178 ( mag ) & 1.11 @xmath170 0.14 & 1.23 @xmath170 0.14 + [ + 0.5ex ] @xmath179 ( mag ) & + [ + 0.5ex ] @xmath180 ( mag ) & + [ + 0.5ex ] distance ( pc ) & + [ + 0.5ex ] parallax ( mas ) & + [ + 0.5ex ] @xmath181 ( ) & @xmath182 & @xmath183 + [ + 1ex ] @xmath184 ( ) & @xmath185 & @xmath186 + [ + 0.5ex ] @xmath19 ( ) & 15 @xmath170 1 & 15 @xmath170 1
the accurate properties derived for permit an interesting comparison with predictions from current stellar evolution theory .
figure [ fig : mistlogg ] shows these determinations in the @xmath18 versus @xmath17 plane along with evolutionary tracks for the measured masses from the mesa isochrones and stellar tracks series ( mist ; * ? ? ?
* ) series , which is based on the modules for experiments in stellar astrophysics package ( mesa ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the metallicity in the models has been set to @xmath8 } = -0.18 $ ] in order to match the observations .
the models are in excellent agreement with the observations at an age of 190 myr ; an isochrone for this age is shown in the figure with a dashed line .
in particular , the temperature difference between the components determined from spectroscopy is consistent with the separation between the evolutionary tracks , which depends on the mass ratio .
the stars are seen to be little evolved from the zero - age main sequence .
the same comparison is shown in the mass - radius and mass - temperature diagrams of figure [ fig : mistmassradius ] , also indicating good agreement .
estimates of the rate of apsidal motion for have varied considerably over the years , beginning with the first measurement by @xcite yielding @xmath187 deg century@xmath10 , later revised to @xmath188 deg century@xmath10 @xcite . a similar estimate of @xmath189 deg century@xmath10 was published by @xcite .
all three studies judged these values to be consistent with expectations from theory ( including classical terms and general relativity ) , although the physical parameters of the components necessary to compute the predicted apsidal motion rate were not particularly well known at the time , as no dynamical masses were available .
much smaller values of @xmath50 were reported subsequently by @xcite , @xmath190 deg century@xmath10 , @xcite , @xmath191 deg century@xmath10 , and @xcite , @xmath192 deg century@xmath10 , all of whom concluded that the apsidal motion of was too slow compared to theory .
finally , @xcite and @xcite reported intermediate values of @xmath193 deg century@xmath10 and @xmath194 deg century@xmath10 , respectively , and considered these to be in good agreement with theory .
the value resulting from our global fit ( table [ tab : results ] ) , @xmath195 deg century@xmath10 , is more in line with the recent studies and is considerably more precise .
it corresponds to an apsidal period of @xmath196 years .
the measurements and the computed ephemeris curve are shown in figure [ fig : ephemeris ] . with the accurate knowledge we now have of the physical properties of the components
, we estimate the predicted rate of periastron advance from classical terms ( tidal and rotational distortions ) to be @xmath197 deg century@xmath10 , where we have adopted identical internal structure constants for the two stars of @xmath198 from the models by @xcite , for the average mass and age of the system . the general relativistic contribution ( e.g. , * ? ? ?
* ; * ? ? ?
* ) is calculated to be @xmath199 deg century@xmath10 , which is 7.5 times larger than the classical effect .
the total expected apsidal motion is then @xmath200 deg century@xmath10 .
this agrees with our measurement within the uncertainties ( at the 0.8@xmath201 level ) .
our measurements of the properties of represent an improvement in the precision of the masses of a factor of @xmath215 over the work of @xcite , and a factor of @xmath213 in the radii .
much of this is due to the greater number and higher resolution of our spectroscopic observations , and partly also to the additional light curves from kelt .
the absolute dimensions of the system now rank among the best for eclipsing binaries ( see , e.g. , * ? ? ?
the agreement with stellar evolution theory is excellent , and suggests an age of 190 myr according to the mist models .
the abundance we infer from this comparison , @xmath8 } = -0.18 $ ] , is not out of the ordinary .
the @xmath202 space motion components based on our center - of - mass velocity ( cfa zero point ) , distance , and the proper motion from the first data release ( dr1 ) of the _ gaia _ mission @xcite are @xmath203 , @xmath204 , and @xmath205 in the lsr frame .
is counted positive toward the galactic center . ]
these are typical of the thin disk , as expected from the youth of the system .
is a member of a very small group of eclipsing systems with well determined properties and accurately measured apsidal motion in which the contribution from general relativity is significant . in this particular example
the gr effect represents 88% of the total predicted motion .
estimates of @xmath50 over the last 30 years have relied almost exclusively on measurements of times of minimum light , and have varied by nearly a factor of two depending largely on the sample of timings used .
in addition to having an updated list of historical timings going back to 1958 , in this work we have added to the analysis radial velocities spanning 22 years and light curves separated in time by about 30 years , all of which provide additional information constraining the rate of periastron advance . as a result ,
the formal uncertainty of @xmath50 has been reduced significantly to better than 5% .
we find that the observed rate agrees with the predicted rate within the errors , consistent with some of the more recent but less precise estimates from the last 15 years .
even so , it would be of considerable interest to attempt a determination of the orientation of the spin axes of the components through the rossiter - mclaughlin effect , which was used successfully in di her to show that the axes of those stars are tilted relative to the orbit , explaining the anomalously slow apsidal motion of that system that had puzzled astronomers for decades . since there does not appear to be any disagreement with the expected value of @xmath50 for
, we would not expect the spin axes to be strongly misaligned .
our accurate distance estimate for leads to an inferred parallax of @xmath206 mas that is marginally consistent with the trigonometric value of @xmath207 mas from the first data release ( dr1 ) of the _ gaia _ mission , where the uncertainty quoted in the last value does not include an estimated contribution of 0.30 mas from systematic errors @xcite .
recent studies @xcite have indicated there may in fact be a small but significant bias in the _ gaia _ parallaxes of about 0.25 mas , in the sense that the original _ gaia _ values are too small .
we note that this happens to go in the direction of bringing agreement in the case of
. finally , a detailed abundance analysis of would be highly beneficial as it would permit a more stringent comparison with stellar evolution models than performed here . because a change in abundance shifts the evolutionary tracks horizontally in figure [ fig : mistlogg ] , knowing the metallicity could serve to test the accuracy of the temperature estimates from this work .
we are grateful to p. berlind , m. calkins , r. j. davis , d. w. latham , and r. p. stefanik for help in obtaining the spectroscopic observations of , and to r. j. davis and j. mink for maintaining the cfa echelle database over the years .
we also thank j. irwin for helpful discussions about the use of his light - curve code and for implementing the apsidal motion capability at our request , a. claret for providing the stellar evolution tracks used for the apsidal motion calculation , and the anonymous referee for a helpful comment .
g.t . acknowledges partial support for this work from nsf grant ast-1509375 .
the efforts of c.m . were supported by the sao reu program , funded in part by the national science foundation reu and department of defense assure programs under nsf grant no . 1262851 , and by the smithsonian institution .
work performed by j.e.r . was supported by the harvard future faculty leaders postdoctoral fellowship .
this research has made use of the simbad and vizier databases , operated at cds , strasbourg , france , and of nasa s astrophysics data system abstract service .
|
we report new spectroscopic and photometric observations of the main - sequence , detached , eccentric , double - lined eclipsing binary ( @xmath0 days , @xmath1 ) . using these observations together with existing measurements we determine the component masses and radii to better than 1% precision : @xmath2 , @xmath3 , @xmath4 , and @xmath5 .
the nearly identical b9.5 stars have estimated temperatures of @xmath6 k and @xmath7 k. a comparison of these properties with current stellar evolution models shows excellent agreement at an age of about 190 myr and @xmath8 } \approx -0.18 $ ] .
both components are found to be rotating at the pseudo - synchronous rate .
the system displays a slow periastron advance that is dominated by general relativity ( gr ) , and has previously been claimed to be slower than predicted by theory .
our new measurement , @xmath9 deg century@xmath10 , has an 88% contribution from gr and agrees with the expected rate within the uncertainties .
we also clarify the use of the gravity darkening coefficients in the light - curve fitting program ebop , a version of which we use here .
|
in this paper we consider the spaces @xmath3 of imbeddings of the circle into @xmath2 and the spaces @xmath6 of imbeddings of the real line into @xmath2 with fixed behavior at infinity , namely , imbeddings that coincide with a fixed imbedded line in @xmath2 outside a compact subset .
when @xmath7 one can define , using configuration space integrals , chain maps from certain graph complexes @xmath8 or @xmath9 to the de rham complexes of the spaces of imbeddings : @xmath10 we will use the same symbol @xmath11 also for the induced maps in cohomology .
our main interest is to determine which algebraic structures are preserved by the maps @xmath11 . on the one side , in fact ,
the de rham complexes are differential graded commutative algebras .
moreover , there exist a multiplication @xmath12 given by attaching the end of the first imbedding to the beginning of the second , and rescaling .
this operation is non - associative but gives rise , together with the cup product , to a hopf algebra structure on @xmath13 for @xmath7 . indeed , as shown in @xcite , the multiplication is part of an action of the operad of little 2-cubes on @xmath6 . on the other side ,
we show in section [ sec : alg ] that @xmath8 and @xmath9 ( whose definitions are recalled in section [ grcoho ] ) can be endowed , respectively , with a differential algebra and a differential hopf algebra structure .
the existence of these algebraic structures implies in particular the existence of infinitely many nontrivial classes of trivalent cocycles of even type ( see section [ sec : inf ] ) , and , as a consequence of the results of @xcite , of infinitely many nontrivial elements of @xmath14 , for every even @xmath15 .
the central result of the paper , contained in section [ sec : hopf ] , is that , for @xmath7 , the maps @xmath11 in cohomology respect all the above algebraic structures .
finally , in section [ sec : anomaly ] we consider the case @xmath16 and @xmath17 ( the case of vassiliev invariants ) and we give a new interpretation of the so - called `` anomaly '' term @xcite as the obstruction for the bott
taubes map from trivalent cocycles to vassiliev invariants to be a coalgebra map . as a final remark
, we recall that there exist other approaches to study the cohomology of spaces of imbeddings based on graph cohomology @xcite
. it would be interesting to understand our results in these other contexts .
we thank domenico fiorenza , dev sinha , jim stasheff , victor tourtchine and the referee for useful comments and suggestions on a first version of this paper .
a. s. c. acknowledges the university of roma `` la sapienza '' ( special trimester `` moduli , lie theory , interactions with physics '' ) , r. l. and p. c .- r .
acknowledge the university of zurich , r. l. acknowledges the university of milano for their kind hospitality during the preparation of the work .
we briefly recall the definition of the _ graph complexes _ given in @xcite , section 4 .
a graph consists of an _ oriented circle _ and many _ edges _ joining vertices .
the vertices lying on the circle are called _ external vertices _ , those lying off the circle are called _ internal vertices _ and are required to be at least trivalent .
we define the order @xmath18 of a graph to be minus its euler characteristic , and the degree @xmath19 to be the deviation of the graph from being trivalent , namely : @xmath20 where @xmath21 is the number of edges , @xmath22 the number of internal vertices and @xmath23 the number of external vertices of the graph .
the graph complexes @xmath8 depend only on the parity of @xmath4 , and we will denote by @xmath24 and @xmath25 the real vector spaces generated by decorated graphs of order @xmath26 and degree @xmath27 , of odd and even type , respectively .
the type of a graph depends on the decoration which we put on it . by definition ,
the decoration in @xmath24 is given by numbering all the vertices ( with the convention that we first number the external vertices and then the internal ones ) and orienting the edges .
then one takes the quotient by the following relations : a cyclic permutations of the external vertices or a permutation of the internal vertices multiplies the graph by the sign of the permutation , a reversal of an orientation of an edge produces a minus sign .
an extra decoration is needed on edges connecting the same external vertex , namely an ordering of the two half - edges forming them .
the decoration in @xmath25 is given by numbering the external vertices and numbering the edges , while the relations are as follows : a cyclic permutations of the external vertices or a permutation of the edges multiplies the graph by the sign of the permutation .
double lines and internal loops are not allowed , namely , in @xmath24 and @xmath25 we quotient by the subspace generated by all the graphs containing two edges joining the same pair of vertices and by all the graphs containing edges whose end - points are the same internal vertex . by an _ arc _
we mean a piece of the oriented circle between two consecutive vertices , and by a _ regular edge _ we mean an edge with at least one internal end - point . the coboundary operators @xmath28 and @xmath29 are linear operators , whose action on a graph @xmath30 is given by the signed sum of all the graphs obtained from @xmath30 by contracting , one at a time , all the regular edges and arcs of the graphs .
the signs are as follows : if we contract an arc or edge connecting the vertex @xmath31 with the vertex @xmath32 , and oriented from @xmath31 to @xmath32 , then the sign is @xmath33 if @xmath34 or @xmath35 if @xmath36 . if we contract an edge labeled by @xmath37 , then the sign is @xmath38 where @xmath23 is the number of external vertices of the graph .
the decoration on @xmath30 induces a decoration on the contracted graphs as follows : contraction of the edge between the vertex @xmath31 and the vertex @xmath32 produces a new vertex labeled by @xmath39 while the vertex labels greater than @xmath40 are rescaled by one .
similarly when the edge labeled by @xmath37 is collapsed , the edge labels greater than @xmath37 are rescaled by one .
the complexes @xmath41 and @xmath42 are defined in the same manner , except that instead of graphs with an oriented circle we consider graphs with an _
oriented line_. the cohomology groups with respect to the above coboundary operators will be denoted by @xmath43 , @xmath44 , @xmath45 and @xmath46 .
when we do not want to specify the parity of these complexes we simply write @xmath47 or @xmath48 .
moreover we set @xmath49 , @xmath50 , @xmath51 and @xmath52
suppose that @xmath53 and @xmath54 are two graphs with @xmath55 and @xmath56 external vertices , respectively .
the sets @xmath57 and @xmath58 of external vertices of @xmath53 and @xmath54 respectively , are cyclically ordered .
a @xmath59-shuffle is a permutation of the set @xmath60 respecting the cyclic order of @xmath57 and @xmath58 .
if @xmath61 is a @xmath59-shuffle , then we can combine @xmath53 and @xmath54 in a single graph @xmath62 with @xmath63 external vertices by placing the legs of @xmath53 and @xmath54 into distinct external vertices on an oriented circle , according to @xmath61 .
we put the labels in @xmath64 as follows : the vertices and edges which come from @xmath53 are numbered in the same way as in @xmath53 .
next we number the vertices coming from @xmath54 by adding to the corresponding label of @xmath54 the number of labeled vertices of @xmath53 and the edges by adding to the corresponding label of @xmath54 the number of edges of @xmath53 .
if a vertex or edge is not labeled in @xmath53 or @xmath54 , it remains unlabeled in @xmath64 the multiplication @xmath65 of two graphs @xmath53 and @xmath54 is then defined as @xmath66 where the sum is taken over all the @xmath59-shuffles and @xmath67 in the above formula , @xmath68 denotes the number of edges of @xmath53 . extending this product to @xmath69 by linearity
, we obtain an associative operation called the _ shuffle product _ : @xmath70 we define a new grading @xmath71 of the graphs generating @xmath72 and @xmath73 , by considering the total number of labels of the graph : @xmath74 where @xmath75 is the total number of vertices of @xmath30 .
modulo 2 , the new grading is equal to @xmath76 in the even case and to @xmath27 in the odd case . with respect to the new grading the integration maps @xmath11
will be grading - preserving modulo 2 .
the shuffle product is graded commutative with respect to this new grading .
in fact one can easily see that @xmath77 .
leibnitz rule also holds between the coboundary operator @xmath78 , which has degree @xmath79 with respect to the new grading , and the shuffle product , which has degree 0 .
more precisely [ prop : db ] @xmath80 for every @xmath81 .
let @xmath61 be a @xmath59-shuffle .
when we apply the coboundary operator @xmath78 to one of the graphs of @xmath64 we can either collapse an edge of @xmath53 , or collapse two external vertices of @xmath53 , or collapse an edge of @xmath54 , or collapse two external vertices of @xmath54 , or collapse an external vertex of @xmath53 with an external vertex of @xmath54 . the first and second contribution yield @xmath82 , while the third and fourth yield @xmath83 . taking into account the signs
we obtain the formula @xmath84 where @xmath85 is a linear combination of the graphs obtained from @xmath64 by contracting an external vertex of @xmath53 with an external vertex of @xmath54 . now
, suppose that @xmath61 brings the @xmath31th external vertex of @xmath53 next to the @xmath32th external vertex of @xmath54 in @xmath62 .
collapsing the arc between these two vertices yields a contribution , called @xmath86 , to @xmath87 .
then consider the @xmath59-shuffle @xmath88 obtained by composing @xmath61 with a transposition of the two vertices considered above .
collapsing the arc in @xmath89 between these vertices yields the contribution @xmath90 to @xmath87 .
this fact in turn implies that @xmath91 , with the sum taken over all @xmath59-shuffles . as a consequence
, @xmath92 is a differential graded commutative algebra .
it also has a unit @xmath93 consisting of the graph without edges .
it follows that its cohomology @xmath94 is a graded commutative algebra with unit .
the product @xmath95 on @xmath48 is defined exactly as in the previous case , i.e. , as the shuffle product of two graphs .
the unit * 1 * is the graph with no edges .
in addition we have a comultiplication @xmath96 mapping a graph @xmath30 to the signed sum of @xmath97 over all possible ways to cut @xmath30 into two disconnected parts @xmath98 and @xmath99 by removing an internal point of one of the arcs of @xmath30 . by convention , if the oriented line is oriented , say , from left to right , then @xmath53 is the left - most component of @xmath100 .
a graph @xmath30 is primitive if it can not be disconnected in a nontrivial way by removing an internal point of an arc , and in this case @xmath101 . to fix the signs in the general case , first of all we order the primitive subgraphs of @xmath30 with respect to the oriented line , and assume that the labels are compatible with this ordering ( i.e. , all labels of a primitive subgraph are less than any label of a subsequent one ) .
the decoration induced by @xmath30 on @xmath98 and @xmath99 is determined by rescaling the labels of @xmath99 and @xmath102 where @xmath103 is given in eq . .
we also define a counit @xmath104 by @xmath105 if @xmath106 and @xmath107 . with the above definitions ,
@xmath108 is a differential graded commutative hopf algebra with unit .
the leibnitz rule and the graded commutativity of the shuffle product hold just as for @xmath69 , and one can easily verify the coassociativity of the coproduct and the compatibility between product and coproduct .
the coboundary operator @xmath78 induces a coboundary operator on @xmath109 by setting @xmath110 on the generators and extending this definition to @xmath111 by linearity .
then one has the following equation @xmath112 in fact , it easy to check that , for any graph @xmath30 , the explicit expression for @xmath113 contains the same graphs ( with the same signs ) of the explicit expression for @xmath114 .
finally , since this bialgebra is @xmath115-graded and the only element in degree zero is the unit , the antipode is automatically defined .
if we pass to the dual , then our algebraic structures correspond to the ones considered in @xcite ( see also @xcite for the degree zero in the odd case ) .
a consequence of the algebra structure on @xmath116 is the following [ nonzerop ] for every @xmath117 we have @xmath118 it is well known that the odd - case analogue of this proposition holds , thanks to the existence of vassiliev knot invariants at any order @xcite .
let us denote by the @xmath119 the graph cocycle of figure [ ec2 ] .
we want to show that , for every @xmath120 , the @xmath121th power of @xmath122 is nontrivial . first , we notice that @xmath123 and hence , thanks to proposition [ prop : db ] , we also have @xmath124
. moreover @xmath125 can not be @xmath126-exact since it has degree zero and there are no graphs of negative degree .
therefore , it is enough to prove that , for every @xmath127 , the cocycle @xmath125 is a linear combination of graphs with at least one coefficient different from zero .
let us denoted by @xmath128 the graph represented in figure [ xyn ] .
this graph has @xmath121 triples of edges are attached on the external circle in such a way that they do not overlap .
we will prove by induction that the coefficient of @xmath128 in @xmath125 is different from zero for every @xmath120 .
the proof of this fact for @xmath129 goes as follows : in the shuffle product @xmath130 there are exactly three contributions to @xmath54 , all coming from @xmath131 . since each of these three contributions has coefficient @xmath132 , we have that the coefficient of @xmath54 in @xmath133 is different from zero .
next , we suppose that the coefficient of @xmath134 in @xmath135 is different from zero , and we observe that all the contributions to the graph @xmath128 in @xmath125 arise from the shuffle product of @xmath134 with @xmath53 .
an easy computation shows that these contributions are all equal up to an even permutation of labels , and hence they can not cancel out .
it is known from @xcite , theorem 1.1 , that the maps @xmath136 are injective for every @xmath7 .
this means that each of the graph cocycles of the above proposition produces a nontrivial cohomology class of @xmath3 for every even @xmath15 .
hence we have : for any @xmath137 and for any positive integer @xmath138 , there are nontrivial cohomology classes on @xmath3 of degree greater than @xmath138 .
we now recall how the maps of eqs . and are constructed ( see @xcite for further details ) .
we consider the fiber bundle @xmath139 whose fiber over a given imbedding @xmath140 is the compactified configuration space of @xmath141 points in @xmath2 , the first @xmath142 of which are constrained on @xmath140 .
let us fix a symmetric volume form @xmath143 on @xmath144 , namely a normalized top form satisfying the additional condition @xmath145 , where @xmath37 is the antipodal map .
a tautological form @xmath146 is by definition the pull - back to @xmath147 of @xmath143 via the smooth map @xmath148
which , on the interior of @xmath147 , is defined as @xmath149 for @xmath150 we use instead the map : @xmath151 where @xmath152 is a component of the compactified configuration space of @xmath142 points on @xmath0 , @xmath153 forgets the @xmath154 points not lying on the imbedding , @xmath155 is the projection on the @xmath31th point and @xmath156 is the normalized derivative @xmath157 . for any given graph @xmath158 with @xmath142 external vertices and @xmath154 internal vertices ,
we construct a differential form @xmath159 on @xmath147 by associating the tautological form @xmath146 to the edge connecting the vertices @xmath31 and @xmath32 , and taking the wedge product of these forms over all the edges of @xmath30 .
then @xmath160 is set to be the integral of @xmath159 along the fibers of @xmath139 .
the map @xmath11 extended by linearity to @xmath47 takes value in @xmath161 .
similarly , one can define @xmath162 .
+ let us now turn to the algebraic structures in the case of imbeddings of @xmath0 into @xmath2 . we know from @xcite , theorem 4.4 , that the integration map @xmath163 is a chain map for @xmath7 .
an easy check shows that the shuffle product correspond exactly to the wedge product of configuration space integrals .
therefore we have : the integration map @xmath164 is a homomorphism of differential algebras with unit for @xmath7 .
next we consider @xmath6 .
[ chl ] the integration map @xmath165 is a chain a map for @xmath7 .
the only difference with respect to the case of imbeddings of the circle is that now one has to consider also faces describing points on the imbedding escaping to infinity , possibly along with external points .
a main feature of the compactified configuration spaces is that they split near the codimension - one faces in a product of the configurations collapsing to a certain point ( or escaping at the point `` infinity '' ) and the configurations which remain far from this collapsing point . since the imbeddings are fixed outside a compact set , the integration along the faces at infinity yields zero unless the form degree of the integrand is zero on the first component ( i.e. , the configurations escaping at infinity ) .
this implies that an entire connected component of a graph has to escape to infinity : in fact , whenever exactly one argument of a tautological form goes to infinity , the form degree is entirely carried by the point escaping at infinity as a consequence of @xmath166 if however a connected subgraph @xmath30 yields a zero form after integration on the face at infinity , the relation @xmath167 should hold ( where again @xmath21 is the number of edges , @xmath23 the number of external vertices and @xmath22 the number of internal vertices of @xmath30 ) . but in a graph whose vertices are at least trivalent @xmath168 is nonnegative .
the proof of the next proposition is exactly as for @xmath3 .
the integration map @xmath169 is a homomorphism of differential algebras with unit for @xmath7 .
we now want to define a coproduct on @xmath170 . in the following we will call _ long knots _ the elements of @xmath6 .
we first fix some convention in the definition of @xmath6 ; namely , we choose a basis @xmath171 of @xmath2 such that all the elements of @xmath6 coincide , outside a compact subset , with the reference imbedding @xmath172
. then we observe that two long knots @xmath173 and @xmath174 can be composed to a new long knot @xmath175 by @xmath176 where @xmath177 and @xmath178 are any pair of diffeomorphisms , e.g. , @xmath179 .
roughly speaking we are attaching @xmath173 and @xmath174 one after the other .
the product @xmath27 has a unit @xmath21 given by the linear imbedding @xmath180 used to define @xmath6 .
the pullback of the product @xmath181 defines a non - coassociative coproduct @xmath182 where @xmath183 denotes the topological tensor product of frchet spaces .
the evaluation at @xmath21 defines a counit @xmath184 : @xmath185 we observe however that the non - coassociative coproduct @xmath186 gives rise to an associative operation in cohomology ( in fact the product @xmath27 is associative up to a homotopy given by composing the long knot with a suitable diffeomorphism of @xmath1 ) .
more precisely , we have : @xmath187 is a graded commutative and cocommutative hopf algebra for @xmath7 .
as shown in @xcite , there exists an action of the little 2-cubes operad @xmath188 on @xmath6 .
this means that there are operations on the space @xmath6 corresponding to each element of @xmath188 .
in particular , it turns out that one of these operations is the multiplication @xmath27 described above . passing on the cochain level we obtain an action of operad @xmath189 on @xmath170 , which give rise in cohomology to an action of the operad @xmath190 on @xmath13 . since the operad @xmath190 is the linear dual of the gerstenhaber operad , we have that the coproduct is coassociative and cocommutative .
the compatibility between the wedge product and the coproduct is obvious .
finally , the existence of the antipode follows from the fact that @xmath13 is @xmath115-graded with only one element in degree zero . a more explicitly proof of the cocommutativity of the coproduct is based on the observation that the composition @xmath27 of two long knots is commutative up to homotopy .
in fact , one can shrink one of the two long knots in a very small region and slide it along the other long knot ( see @xcite for details ) .
our central result is then the following [ main ] the map @xmath191 is a hopf algebra homomorphism for @xmath7 .
first , we notice that for @xmath7 , the degree of the differential form @xmath160 is zero if and only if @xmath192 .
therefore we have @xmath193 .
we now have to show that the coproducts are compatible ; viz .
, @xmath194 to prove this , consider any two cycles @xmath195 and @xmath196 ( of degree @xmath26 and @xmath121 ) of imbeddings .
let @xmath197 be the @xmath198-cycle obtained by attaching the two cycles : @xmath199 identity is equivalent to @xmath200 for any @xmath195 and @xmath196 and for any @xmath201 , where we write @xmath202 . to prove , let us introduce the @xmath203-chain @xmath204 by @xmath205 with @xmath206 . in practice
we are moving the support of the cycles @xmath195 and @xmath196 far apart , and the parameter @xmath207 measure the distance between the two cycles .
since @xmath208 and since @xmath11 is a chain map ( lemma [ chl ] ) , by stokes theorem we get @xmath209 with @xmath210 let us write our graph cocycle as @xmath211 .
each @xmath212 can be split into two parts as follows .
we fix @xmath213 and suppose @xmath214 has @xmath142 external vertices and @xmath154 internal vertices
. then we define @xmath215 to be the subbundle of @xmath147 where the points corresponding to the external vertices of every primitive subgraph of @xmath214 lie either all to the left of @xmath207 or all to the right of @xmath216 .
we also let @xmath217 be the fiberwise complement of @xmath215 in @xmath147 .
now we define @xmath218 , @xmath219 , to be the integral of the differential form @xmath220 performed along the fibers of @xmath221 .
finally we set @xmath222 and we have @xmath223 .
one immediately sees that @xmath224 is equal to the right - hand side of eq . , and
hence what we have to prove is that @xmath225 we now need a generalization of lemma 10 of @xcite .
let @xmath30 be a connected graph with @xmath23 external vertices and @xmath22 internal vertices , and let @xmath159 be the product of tautological forms associated to the graph @xmath30 .
we denote by @xmath226 the integral of @xmath159 over the internal vertices of @xmath30 , namely the push - forward of @xmath159 along the map @xmath227 that forgets the internal points .
let us write @xmath226 in coordinates as @xmath228 , where @xmath11 is a multi - index .
suppose moreover that @xmath229 a sequence in the configuration space with the property that there is a pair of points whose distance diverges as @xmath230 goes to infinity .
[ lemma1 ] with the above notations , we have @xmath231 case 1 .
we consider first the case when @xmath30 has no edges whose end - points are both external .
then @xmath226 turns out to be bounded whenever its arguments @xmath232 run in a bounded subdomain of @xmath2 ( namely , @xmath226 do not diverge if two or more points collapse ) .
we can always suppose that the pair of points whose distance diverges fastest are @xmath233 and @xmath234 .
using the translation invariance of the integral , we also suppose @xmath235 .
therefore we have that @xmath236 as @xmath237 .
we claim that if we rescale all the variables by @xmath238 we get @xmath239 where @xmath240 .
in fact , when performing the integral along the fibers of @xmath227 , it is convenient to rescale the integration variables by @xmath238 .
this yields the contribution @xmath241 . at this point
the tautological forms @xmath146 , whose degree is @xmath242 and whose number is the same as the number of edges @xmath21 , are obtained by rescaling the arguments of the functions @xmath243 , and this yields the contribution @xmath244 . moreover , using the fact that @xmath245 and @xmath246 , we have @xmath247 now , @xmath248 is bounded for every @xmath249 , and hence the quantity @xmath250 remains bounded when @xmath251 goes to infinity .
this proves the first case .
when @xmath30 has edges whose end - points are both external , we denote by @xmath252 the same graph with these edges removed .
we have then @xmath253 where @xmath254 is the product of the functions associated to the removed edges .
if @xmath252 is the empty graph , then there is at least one edge whose end - points go far apart .
since the imbeddings are fixed outside a compact set , asymptotically @xmath254 is the pull - back of a volume form via a constant function , i.e. , @xmath254 vanishes in the limit @xmath237 .
if @xmath252 is non empty but connected we are done by the argument in case 1 .
the last possibility is when @xmath252 is non empty and disconnected .
if inside a connected component there is a pair of points whose distance diverges , we are also done by the argument in case 1 , otherwise there is at least one edge whose end - points go far apart , and in this case @xmath254 goes to zero .
we now show that eq .
suppose @xmath214 has @xmath142 external vertices and @xmath154 internal vertices , and consider @xmath217 .
for every element of @xmath217 consider a primitive subgraph @xmath255 of @xmath214 such that the preimage @xmath121 of its left - most point on the imbedding is less than @xmath216 , while the preimage @xmath256 of its right - most point is greater than @xmath207 .
let @xmath257 be the connected component of @xmath255 minus the oriented line , ordered following the order of their left - most external vertices . with the symbols @xmath258 and @xmath259 we mean the preimages of the left - most and right - most among the points corresponding to the external vertices of @xmath260
. in particular , @xmath261 .
if @xmath262 then the result follows from lemma [ lemma1 ] applied to @xmath263 .
if @xmath264 then @xmath265 . if @xmath266 then we repeat our considerations for @xmath267 , otherwise we have two possibilities : @xmath268 , and the integral of @xmath220 on this portion of @xmath217 is zero for dimensional reasons , or @xmath269 and we can apply lemma [ lemma1 ] to @xmath270 . this concludes the proof of the theorem .
let us consider now the case of ordinary framed long knots , i.e. , imbeddings of @xmath1 into @xmath5 with a choice of a trivialization of their normal bundle .
following @xcite , we know that @xmath11 is not a chain map because of an anomaly term due to contributions from the most hidden faces of the boundaries of the compactified configuration spaces . more precisely , if we restrict to trivalent graphs , the following equation holds @xmath271 where @xmath272 here @xmath273 are certain ( unknown ) coefficients while @xmath274 is the graph with one chord only , so @xmath275 is the self - linking number @xmath276 of the framed long knot @xmath277 . as a consequence , the bott
taubes map @xmath278 associates to each cocycle of trivalent graphs an invariant of framed long knots .
let @xmath30 be a cocycle of trivalent graphs , @xmath195 and @xmath196 two framed long knots , @xmath279 their product and @xmath204 the 1-chain of eq . .
then @xmath280 and hence , applying stokes theorem , we have @xmath281 since the self - linking number of @xmath282 does not depend on @xmath207 and it is therefore equal to the self - linking number of @xmath283 . as a consequence , using the same arguments of the proof of theorem [ main ] , we deduce that also in three dimensions the map @xmath11 , when restricted to trivalent graphs , is compatible with the coproducts . on the other hand the bott
taubes map @xmath278 is a map of coalgebras if and only if @xmath284 the left - hand side of is given by @xmath285 while the right - hand side is @xmath286 in particular we can write eq . in the case
when @xmath30 is obtained by `` attaching '' the oriented lines of two graphs @xmath53 and @xmath54 . by using the fact of @xcite that @xmath287 if @xmath30 is not primitive , then one easily sees that holds if and only if @xmath288 hence , if @xmath289 is different from zero , @xmath278 is not a coalgebra map ( e.g. choose @xmath195 , @xmath196 and @xmath54 such that @xmath290 , @xmath291 and @xmath292 ) . in other words ,
the anomaly term @xmath293 is the obstruction for @xmath278 to be a coalgebra map .
|
we define algebraic structures on graph cohomology and prove that they correspond to algebraic structures on the cohomology of the spaces of imbeddings of @xmath0 or @xmath1 into @xmath2 . as a corollary , we deduce the existence of an infinite number of nontrivial cohomology classes in @xmath3 when @xmath4 is even and greater than 3 .
finally , we give a new interpretation of the anomaly term for the vassiliev invariants in @xmath5 .
|
weak measurements as an extension of quantum measurements were first introduced by aharonov , albert , and vaidman @xcite . in quantum measurements , the observable of system couples a probe state with a pointer whose value can be read out by a meter . in general ,
the conventional quantum measurements involve in a process of strong coupling with the probe wavefunction is distorted .
weak measurements suggest that the coupling between the observable and the probe state is weak and the probe wavefunction remains almost undisturbed .
the weak value of an observable @xmath0 outside the eigenvalue spectrum can be obtained and the results are much larger than any eigenvalues of the quantum system . it is shown that the weak value @xmath1 can be formed as a simple expression @xmath2 in which @xmath3 and @xmath4 are the preselected and postselected states , respectively @xcite . indeed , the weak measurements have become a useful tool for high - precision measurements of small physical parameters , such as single - photon tunneling time @xcite , deflections of light beam @xcite , phase shift @xcite , frequency shift @xcite , single - photon nonlinearity @xcite , high - resolution phase estimation @xcite , and angular rotations @xcite .
in addition , it also assists us in researching fundamental questions of quantum mechanics such as single photon s polarization@xcite , hardy s paradox @xcite , photon trajectories @xcite , heisenberg s uncertainty principle @xcite , quantum polarization state @xcite , direct measurement of the quantum wavefunction @xcite , high - dimensional state vector @xcite , and quantum cheshire cat @xcite . as one of important applications , hosten and kwiat develop a weak measurement to detect a tiny spin - dependent splitting in photonic spin hall effect ( she ) @xcite .
such an effect is attributed to spin - orbit interaction and implied by angular momentum conservation @xcite . in the procedure of weak measurements ,
the quantum system is first preselected as a initial state .
then the observable is very weekly coupled the pointer state .
finally , the pointer position is recorded when the quantum system is postselected in a final state .
weak measurements are valid only in the regime of weak coupling between the observable and the probe state @xcite .
however , it does not always satisfy such condition in some practical situations .
the conventional weak measurements should be modified if the coupling strength is not weak enough @xcite .
in addition , when the @xmath3 and @xmath4 are nearly orthogonal , @xmath5 , the weak value @xmath1 can become arbitrary large .
in fact , the weak value @xmath1 should be modified in this situation @xcite .
note that the probe wavefunction is distorted in these two cases , and neither conventional weak nor strong measurements can detect the spin - dependent splitting in photonic she . in this paper
, we develop a modified weak measurement for detecting photonic she when the probe wavefunction is distorted .
we consider two possible cases leading to the distortion in the process of measurements : one is due to the strong coupling , the other to the preselected and postselected states are nearly orthogonal .
we find that the measuring procedure with preselected and postselected ensembles is still effective when the probe wavefunction is strongly distorted .
the paper is organized as follows : in sec .
ii , both the conventional and the modified weak measurements for detecting the photonic she are established .
subsequently , the evolution of the wavefunction with different preselected and preselected states in the weak measurements is analyzed in detail . in sec .
iii , the experimental and theoretical results are compared and discussed .
in contrast to the value of the conventional theory , our experimental data agree well with the modified theory . in sec .
iv , a summary is given .
in this section , we develop a modified theoretical model of the weak measurements for detecting photonic she . as a comparison , the conventional weak measurements are also reviewed . in quantum system of the weak measurements ,
an initial state is first prepared .
after the system is weakly coupled with measuring device , the observable @xmath0 undergos a separate degree which is interpreted as a meter .
then we read out the information from it when the postselected state is performed . here , the transverse spatial distribution of light is used as a meter and the observable is @xmath6 . for simplicity , we only consider the preselected states @xmath7 and @xmath8 . with spin basis @xmath9 and @xmath10
, we have the expressions @xmath11 . as an example
, we consider the photonic she in reflection at air - glass interface .
to start , in our weak measurements , the initial state @xmath7 is first preselected .
we only consider the packet spatial extent in @xmath12 direction , and the total wavefunction can be written as @xmath13 where @xmath14 is the fourier transform of @xmath15 .
we assume that @xmath14 is gaussian spatial distribution here . at the interface the light beam separates into two wave packets of orthogonal spin states @xcite @xmath16/{\sqrt{2}}\label{asl},\end{aligned}\ ] ] where @xmath17 is given by @xmath18 here , @xmath19 and @xmath20 represent the fresnel reflection coefficients for parallel and perpendicular polarizations , respectively .
@xmath21 denotes the incident angle and @xmath22 is the rayleigh length .
taking the interaction hamiltonian @xmath23 into account on reflection , the initial state becomes @xmath24 in the weak measurements of photonic she , the meter states corresponding to observable states @xmath9 and @xmath10 remain overlap .
that is @xmath25 , in which @xmath26 is the width of the wavefunction . under such condition , we expand the operator @xmath27 as @xmath28 .
therefore , @xmath29 with the relation of eqs .
( [ asj ] ) and ( [ asu ] ) , the meter state after postselection evolves as @xmath30\nonumber\\ & = \langle\psi_{f}|\psi_{i}\rangle\int d k_{y } \phiup(k_{y})|k_{y}\rangle[1-ik_{y}\delta^{h } a_w]\nonumber\\ & \approx\langle\psi_{f}|\psi_{i}\rangle\int d k_{y } \phiup(k_{y})|k_{y}\rangle e^{-ik_{y}\delta^{h } a_w}\nonumber\\ & = \langle\psi_{f}|\psi_{i}\rangle\int d y \psi(y-\delta^{h } a_w)|y\rangle\label{ass},\end{aligned}\ ] ] where @xmath31 is the conventional formalism of the weak value which is the same as eq .
( [ asi ] ) . from the restrictions pointed out in @xcite , the validity of above calculation requires @xmath32 and @xmath33 the preselected state @xmath3 here is the pure polarization state @xmath7 and the postselected state is @xmath34 , in which @xmath35 is referred to as the postselected angle .
that is @xmath36 in the spin basis , they become @xmath37 the operator @xmath0 between these two states is @xmath38 since we deal with the left- and right - handed circularly polarization basis . by calculating the matrix elements
, we obtain the weak value : @xmath39 and from eqs .
( [ wvc ] ) and ( [ wvq ] ) , the results is valid if @xmath40\label{tlq}.\ ] ] in general , the weak value is a complex number of which the real and imaginary parts correspond to the shifts of the position and momentum in the wavefunction , respectively @xcite .
it is manifested experimentally in the plasmonic spin hall effect @xcite . here ,
the pure imaginary weak value @xmath41 converts the position displacements @xmath17 into a momentum shift @xcite .
the significance of weak value is also attempted to be further understood in recent works @xcite . we next consider the free evolution of the wavefunction before detection .
the displacement of the meter can be describe as @xmath42 in which the factor @xmath43 depends on the meter state and free evolution @xcite . at any given plane @xmath44 ,
the free evolution factor is given by @xmath45 , so we finally obtain the amplified shift of the conventional theory as @xmath46 where @xmath47 is also defined as the conventional amplified factor here .
equation ( [ asz ] ) suggests that in conventional theory , the amplified shift , as well as the amplified factor @xmath47 , is proportional to the absolute value of the weak value .
( color online ) schematic of quantum measurements with preselected and postselected ensembles when probe wavefunction is undisturbed ( a ) or distorted ( b ) and ( c ) .
system is preselected an initial state @xmath3 and its wavefunction exhibits a gaussian profile .
after weak coupling with the system , the wavefunction splits into two spin components(red dotted line and blue dotted line , respectively ) . @xmath48 and
@xmath49 represent the transverse shifts of left- and right - circularly polarized components , respectively .
then we get the final pointer position proportional to @xmath50 after postselection .
the green solid lines indicate the total wave shapes .
( a ) the profile of wavefunction is always gaussian in the procedure of weak measurement .
the conventional weak measurements are valid in this case .
( b ) the profile of probe wavefunction is distorted in strong coupling , where the displacement exhibits a magnitude on the order of the beam width @xmath51 .
( c ) the other case of the distortion of the probe wavefunction .
it occurs when the preselected and postselected states are nearly orthogonal .
the conventional weak measurements are invalid and the modified theory should be developed in the latter two cases.,width=321 ] we now consider the preselected state of @xmath8 . in a similar way we get the same weak value of the observable @xmath41 .
so we have the amplified factor of the initial state @xmath8 as @xmath52 in this case , the original transverse shift is @xcite @xmath53 and the final amplified shift after free propagation is obtained as @xmath54 from the above analysis , it should be noted that the conditions to obtain the weak value of conventional formalism are too strict . in the experiment of the photonic
she , if preselected and postselected states are nearly orthogonal , indicating @xmath5 , the @xmath41 is very large .
thus , the approximations in eq .
( [ ass ] ) is invalid : @xmath55\nthickapprox e^{-ik_{y}\delta^{h , v } a_w}$ ] . on the other hand ,
supposing that the preselected state @xmath7 is incident near the brewster angle , the operator @xmath27 can not be expanded as @xmath28 due to the strong coupling @xcite .
therefore , the weak value @xmath41 in eq .
( [ ass ] ) is inaccurate if one of the conditions is not satisfied .
in fact , the two approximations above require the restriction @xmath56 from eq .
( [ tlq ] ) .
we next calculate the final state of the meter in eq .
( [ ass ] ) to second order : @xmath57 where @xmath58 is the second - order weak value and @xmath59 is the transverse shift @xmath17 or @xmath60 .
the expectation value of the position is written as @xmath61 with the property @xmath62 . here
, the @xmath41 is imaginary and therefore we consider the particular form of @xmath14 with the effective propagation distance @xmath44 : @xmath63 @xcite .
then we can recalculate the position of second - order theory as @xmath64 in the following , we consider the modified theory without approximation .
the measuring procedure with preselected and postselected ensembles is still effective in the modified weak measurements .
for the preselected state @xmath7 , the wave vector @xmath65 should be reconsidered when the wave packet incident near the brewster angle .
the evolution in the state after reflection can be written as @xmath66\nonumber\\ & = |k_{x , y}\rangle\sum\limits_{+,-}\frac{1}{{\sqrt{2 } } } \left(r_{p}-\frac{k_{x}}{k_{0}}\chi \pm ik_{y}r_{p}\delta^{h}\right)|\pm\rangle\nonumber\\ & \rightarrow|y\rangle\sum\limits_{+,-}(y\mp\delta^{h_{mod}})|\pm\rangle\label{esar},\end{aligned}\ ] ] where @xmath67 is defined as the original transverse shift and is given by @xmath68 and @xmath69 .
so the meter state can be described as @xmath70 subsequently , combine eq .
( [ asm ] ) with eq .
( [ mstate ] ) , the final meter state becomes @xmath71\label{fms}.\end{aligned}\ ] ] the expectation value of the pointer observable @xmath12 , also referred to as modified amplified shift , is obtained as @xmath72\sin(2\delta)\cot\theta_{i}}{2k_{0}z_{r}(r_{p}+r_{s})^{2}\cos^{2 } \delta\cot^{2}\theta_{i}+4k_{0}^{2}r_{p}^{2}z_{r}^{2}\sin^{2}\delta}\label{asf}.\end{aligned}\ ] ] as the preselected state @xmath8 , because the coupling is always weak , it still becomes @xmath73/{\sqrt{2}}\label{kyv}.\end{aligned}\ ] ] with the similar calculation in eqs .
( [ mstate])-([asf ] ) , we get the @xmath74 as @xmath75 here , @xmath76 and @xmath77 are the modified amplified factors of states @xmath7 and @xmath8 , respectively
. equations ( [ asf ] ) and ( [ asd ] ) imply that the theoretical output value is no longer proportional to the weak value , which is different from the conventional theory .
but it is worth remarking that the results of the modified theory can reduce to the conventional results as if the condition of weak coupling is satisfied . for the preselected state @xmath8 , when the postselected angle @xmath35 is not too small , the term @xmath78 in eq .
( [ asd ] ) can be ignorable , and eq .
( [ asd ] ) returns to eq .
( [ asp ] ) .
but for the case of @xmath7 , except for the postselected angle limit , the neglect of the term @xmath79 in eq .
( [ asf ] ) requires that the incident angle is far from the brewster angle . as a result ,
( [ asf ] ) can be reduced to eq .
( [ asz ] ) ( under such condition , the @xmath80 can be ignorable ) .
similarly , such analysis also holds for simplifying the amplified factors of the two different theories .
additionally , we point out that the modified weak measurements are also needed for detecting the photonic she with an arbitrary linearly polarized state . if the condition of weak coupling is satisfied , a weak measurement of the photonic she performs as fig .
[ fig1](a ) .
the distribution of the wavefunction is always the gaussian shape during the procedure of weak measurement .
now we consider the preselected state is @xmath7 with incident angle near the brewster angle .
the separation between two spin components becomes large , which causes the distortion of the wavefunction [ fig .
[ fig1](b ) ] . on the other hand , in fig .
[ fig1](c ) , when the preselected and postselected states are nearly or exactly orthogonal , the distortion also occurs after postselection .
the distortion of the prove wavefunction accompanies the violation of the limited condition because in the conventional theory where the weak value is calculated under the assumption that wavefunction remains gaussian .
hence , if the distortion occurs , the quantum system is in the regime where at least one of the two conditions above is violated , and the conventional theory is invalid . to verify it
, we measure the intensity of the beam in our weak measurement for some special cases .
our experimental results are also compared with the conventional theory during the discussion .
the modified theory is valid not only in weak - coupling regime but also in the strong - coupling regime , and especially in the intermediate regime .
to perform the weak measurements of the photonic she , the coordinate frame and the experimental setup are similar to that in ref .
a incident gauss beam is generated by a he - ne laser .
the two nearly crossed polarizers are used to select the initial state @xmath3 and final state @xmath4 . in the experiment
, we chose the preselected state as @xmath7 or @xmath8 , and the corresponding postselected state is @xmath81 or @xmath82 .
the two lenses are used to focus and collimate the beam .
when the light beam impinges on air - glass interface , the tiny spin - dependent splitting takes place . after the light passes through the second lens ,
the amplified shift is detected by a charge - coupled device ( ccd ) .
( color online ) amplified shifts vary with incidence angles .
( a ) and ( b ) show the amplified shifts for preselected state @xmath7 with the fixed postselected angles @xmath35=@xmath83 and @xmath84 , respectively .
( c ) and ( d ) are the cases of @xmath8 with @xmath85 and @xmath86 , respectively .
the solid lines are the modified values by eqs .
( [ asf ] ) and ( [ asd ] ) .
in contrast , the long - dashed lines are the predictions of conventional theory by eqs .
( [ asz ] ) and ( [ asp ] ) , which have no upper bound for preselected state @xmath7 near the brewster angle .
and the short - dashed lines are the result of second - order theory from eq .
( [ sec ] ) .
insets in ( a ) and ( b ) show the difference of the three theories in strong - coupling regime . the experimental data represented by hollow points are also given .
the arrows in the fig .
[ fig2](a ) indicate the two cases in which we measure output intensity of the light beams in fig .
[ fig4].,width=302 ] we first consider the condition of weak coupling . with a fixed preselected angle ,
the coupling strength varies with incident angles . and
then , with the invariable coupling strength , we analyze the other case where the preselected and postselected states are nearly orthogonal .
that is , the incident angle is fixed and the outcome is shown as a function of postselected angles .
we now experimentally measure the amplified shift , amplified factor , and original displacement of the left - circularly polarized component .
we measure the amplified shifts in the case of incident angles varying from @xmath87 to @xmath88 , as shown in fig .
[ fig2 ] . in each case , the experimental results are also given and agree well with the theoretical curves of modified values .
for comparison , the conventional and second - order theories are shown as dashed lines .
for the preselected state @xmath7 [ figs .
[ fig2](a ) and [ fig2](b ) ] , the conventional values are close to the modified values under the condition that the incident angles are far from the brewster angle .
but for the incident angles near the brewster angle , due to the enhancement of spin - orbit interaction , the splitting of two spin components is nearly the same scale as the width of the gaussian beam .
it means that the weak - coupling condition is violated and the gaussian profile of wavefunction is distorted .
remarkably , in the strong - coupling regime , the second - order theory also exhibits large distinction with the modified theory .
( color online ) amplified factors vary with incidence angles .
( a ) and ( b ) show the cases of preselected state @xmath7 .
( c ) and ( d ) are the preselected state @xmath8 . for all panels ,
the long- and short - dashed curves are the conventional and second - order theory from eqs .
( [ awvc ] ) and eqs .
( [ sec ] ) , respectively .
the solid curves are modified values from eqs .
( [ asf ] ) and ( [ asd ] ) .
the experimental data are represented by hollow points.,width=302 ] it is interesting note that for different postselected angles , the divergence between the conventional and modified theories is also different .
it indicates that the condition of postselected angle is also important in the weak measurements as well as the weak - coupling condition . in the case of preselected state @xmath8 with the certain angle @xmath35 , the amplifying shift varies with incident angles as shown in fig .
[ fig2](c ) , both the modified values and the conventional values agree well with the experimental results . because there is not special angle like the brewster angle for preselected state @xmath8 , the interaction between the observable and the probe state is so weak that the weak - coupling condition is always satisfied .
therefore , as long as the angles @xmath35 are not too small , the conventional theory can be approximatively equivalent to the modified theory .
otherwise , the two theories diverge and there is almost no overlap between the modified and conventional curves as shown in fig .
[ fig2](d ) .
( color online ) distortion of wavefunction due to the strong coupling .
the initial state is preselected as @xmath7 , the incident angles as @xmath89 , and @xmath90 , respectively .
in addition , the postselected angles are both chosen as @xmath91 . ( a ) and ( c ) are the theoretical prediction .
( b ) and ( d ) are the experimental results .
the insets indicate the intensity profiles along the @xmath12-axis .
note that , in the regime of strong coupling @xmath90 , the gaussian profile of wavefunction distorts and exhibits asymmetric double - peak .
but the wavefunction still remains gaussian form with @xmath92.,width=302 ] to discuss the problem in detail , the corresponding amplified factors shown in fig .
[ fig3 ] are also given .
generally , according to eq .
( [ awvc ] ) , the amplified factor of the weak measurements is a constant if the postselected angle @xmath35 is decided ( the straight dashed lines ) .
it is valid if two limited conditions are all satisfied . but the modified amplified factors @xmath76 and @xmath77 from eqs .
( [ asf ] ) and ( [ asd ] ) are not constants which are shown as solid curves , which are identical to the curves of second - order theory . for state @xmath7 ,
the amplified factor of the modified theory is consistent with that in conventional theory when the incident angles are far away from the brewster angle , but it becomes small near the brewster angle . as a result , the largest divergence between modified and conventional values occurs .
comparing fig .
[ fig3](a ) with fig .
[ fig3](b ) , we find that the postselected angle has also important impact on the deviation of the two theories .
as predicted , the modified cures coincide well with our experimental data in the strong - coupling regime . for the preselected state @xmath8 , we only concern on the limit of postselected angle . in fig .
[ fig3](c ) , the conventional amplified factor approach to the modified one when the angle @xmath35 is chosen as @xmath84 .
but at first glance , one may regard that there exists discrepancy between the theory and measured data . as a matter of fact ,
such deviation is appropriate because in this panel the spacing of @xmath12-axis is suitably small in order to present the divergence of the two theories .
figure [ fig3](d ) shows the distinction between the two theories with the @xmath93 which is small enough , while our experimental results guarantee the validity of the modified theory .
( color online ) with the fixed incident angles @xmath94 , @xmath95 , and @xmath96 , amplified pointer shifts are shown as functions of postselected angle @xmath35 for preselected states @xmath7 ( left column ) and @xmath8 ( right column ) . here , the modified theoretical values are represented by solid lines from eqs .
( [ asf ] ) and ( [ asd ] ) , and the values in conventional theory are dashed lines from eqs .
( [ asz ] ) and ( [ asp ] ) .
the sharp value of the dashed lines is not shown in each panel due to its infiniteness .
the experimental data are obtained as hollow points .
the arrows in the fig .
[ fig5](f ) are two special cases in which we measure their output intensity shown in fig .
[ fig7].,width=302 ] if the profile of gaussian wavefunction is distorted , the theory of weak measurements should be modified .
there are two possible cases leading to the distortion in the process of measurements : one is due to the strong coupling , the other to the preselected and postselected states are nearly orthogonal .
we first consider the former case .
the initial state is preselected as @xmath7 as shown in fig .
[ fig4 ] , which explains the connection between distortion of the beam and the weak - interaction condition . to avoid the limited condition of postselected angle
, we set it to be @xmath91 .
we consider two special examples labeled by arrows in fig .
[ fig2](a ) .
the measured intensity is read out from ccd ( right column of fig .
[ fig4 ] ) . for comparison ,
the corresponding predictions are also given ( left column of fig .
[ fig4 ] ) .
the case with incident angle @xmath89 suggests that the conventional theory is equivalent to the modified one if the output beam remains the gaussian profile .
conversely , at the incident angle @xmath97 the wavefunction distorts due to the strong coupling and the conventional theory of weak measurements is invalid .
( color online ) amplified factors with various postselected angle .
( a ) , ( c ) , and ( e ) show the cases of @xmath7 .
( b ) , ( d ) , and ( f ) are @xmath8 . in term of eqs .
( [ awvc ] ) , ( [ asf ] ) , and ( [ asd ] ) , the predictions of the modified and conventional theory are represented by solid lines and dashed lines , respectively .
the conventional amplified factor can be arbitrarily large with the angle @xmath35 decreasing to @xmath98 .
the hollow points are experimental data.,width=302 ] all discussion above is about the weak - interaction condition in the modified weak measurements . in the rest of the paper ,
we consider another limited condition of small postselected angle , that is to say the preselected and postselected states are nearly orthogonal .
we find that the weak measurements of conventional theory are also not valid when the postselected angle @xmath35 is too small . to prove that , we detect the photonic she with the fixed incident angles but various postselected angles . in the same sequence , we first measure the amplified shifts with the angle @xmath35 varying from @xmath98 to @xmath99 , which are shown in fig
under such condition , large divergence takes place between modified and conventional theories when the angle @xmath35 is close to @xmath98 . but
the amplified values in conventional theory essentially have no difference to that in modified theory with appropriate postselected angles . in the conventional theory of the weak measurements , according to the eqs .
( [ asz ] ) and ( [ asp ] ) , the amplified value can be arbitrarily large if the preselected and postselected states are nearly crossed , and such value is shown as the dashed lines in the figures .
( color online ) distortion of probe wavefunction due to the fact that the preselected and postselected states are nearly orthogonal .
we chose the preselected state as @xmath8 with a fixed incident angle at @xmath100 .
( a ) and ( c ) are the theoretical results .
( b ) and ( d ) are the experimental results .
the insets : the profiles of wavefunction at the plane of @xmath101 .
it implies that the output wavefunction no longer remains a gaussian profile when the preselected and postselected states are nearly orthogonal.,width=302 ] whereas in our modified theory ( the solid lines ) , with the angle @xmath35 continuously decreasing to @xmath98 , the amplified shift first increases and reaches the maximum value , then decreases rapidly even to @xmath98 .
we do the experiment with the incident angles @xmath94 , @xmath95 , and @xmath96 both for states @xmath7 and @xmath8 , and our experimental data reveal the trend like the modified theory .
note that in fig .
[ fig5](c ) , the modified and conventional curves begin to separate when the @xmath35 is near @xmath99 , but in other cases , such separation takes place until the @xmath35 becomes smaller .
the point is that in this case the weak - interaction is a little strong at the incident angle @xmath102 . to clarify it
, we see that the example in fig .
[ fig5](d ) , with the same incident angle but different preselected state @xmath8 , dose not exist such problem .
we point out that when the postselected angles are negative , the values of amplified shift are the same magnitude but opposite sign , besides , the amplified shift of second - order theory is completely identical to the modified one , which both are not shown here .
( color online ) original transverse shifts of the spin component @xmath9 for preselected states : ( a ) @xmath7 and ( b ) @xmath8 .
the theoretical values are represented by black solid lines from eqs .
( [ as ] ) and ( [ abj ] ) . for each preselected state , the experiment data with two different postselected angles @xmath35 are shown as hollow points .
inset in ( a ) shows the theoretical and experimental results in strong - coupling regime.,width=302 ] in addition , the corresponding amplified factors are also obtained in fig .
we get the conventional amplified factors represented by dashed lines in terms of the relation @xmath103 .
it is shown that the amplified factor has no upper bound when the @xmath35 approaches to @xmath98 .
in fact , both for two preselected states @xmath7 and @xmath8 , the behavior of the amplified factors @xmath76 and @xmath77 shown as the solid lines is similar to that of amplified shifts . in each panel , the modified theory is compared with the conventional theory , and once again , the difference between conventional and modified values would be significant if the preselected angle @xmath35 is small .
the deviation of the amplified factors between the two theories is also determined by incident angle , especially for @xmath7 .
the experimental results agree well with the modified values .
there exists a peak value with an optimal postselected angle , and the amplification effect disappears when the preselected and postselected states are completely orthogonal .
that is to say , by adjusting the postselected angle @xmath35 , one can obtain the maximum amplified factor and improve the precision for measuring the photonic she @xcite .
note that the signal amplification from weak measurements has been extensively studied recently , such as optimal probe wavefunction of weak - value amplification @xcite , technical advantages for weak - value amplification @xcite , and maximizing the output by weak values and weak coupling @xcite .
note that there are still some open questions about whether the weak value ampliation can suppress technical noise @xcite .
we have discussed the connection between the condition of weak - interaction and the output intensity of the light in fig .
[ fig4 ] . here
, we analyze the distortion associated with the postselected angle . in the case of fig .
[ fig5](f ) , we measured its intensity with a decreasing angle @xmath35 , and found that the intensity changes gradually from a single gaussian into asymmetrical double - peak intensity .
then the double - peak intensity becomes symmetric when postselected angle is equal to @xmath98 .
in particular , in fig .
[ fig7 ] , we present the intensity with two postselected angles which are labeled by the arrows in fig .
[ fig5](f ) . both for the predicted and measured intensity
, they remain the gaussian form with the @xmath85 , but distort with the @xmath104 .
we draw a conclusion that with different postselected angles , the experimental data do not fit to the conventional theory when the profile of wavefunction distorts . therefore , one may judge weather the modified theory is equivalent to the conventional one by observing the output intensity of the light . the tiny original shifts of the component of @xmath9 obtained by the modified theory for preselected states @xmath7 and @xmath8 are presented in fig .
[ fig8 ] . for each preselected states
, we detect it with two different postselected angles .
we find that the original shift of @xmath7 is indeed very large near the brewster angle : nearly the same scale as the width of probe state @xmath105 , as the result , the profile is strongly distorted and the weak measurements approximations fail . for the preselected state @xmath8 , the original shifts for various incident angles are always much less than the width of pointer state @xmath106 .
both experimental results agree well with the theoretical predictions .
it should be noted in the strong - coupling case the spin - dependent splitting is very sensitive to the variation of physical parameters and therefore has important applications in precision metrology , such as measuring thickness of metal film @xcite , identifying graphene layers @xcite , determining the strength of axion coupling in topological insulators @xcite , and detecting of magneto - optical constant of magneto - optical media @xcite .
however , in this regime both conventional weak measurements theory and its second - order corrections can not obtain the exact meter shifts as the analysis above .
hence the modified weak measurement is important in precision metrology .
finally , it should be mentioned that photonic she manifests as spin - dependent splitting of light , which corresponds to two types of geometric phases : the rytov - vladimirskii - berry phase associated with the evolution of the propagation direction of light and the pancharatnam - berry phase related to the manipulation with the polarization state of light @xcite . in general , the spin - dependent splitting due to the rytov - vladimirskii - berry phase is limited by a fraction of the wavelength , and can only be detected by weak measurements @xcite .
however , the spin - dependent splitting due to the pancharatnam - berry phase can be large enough for direct detection ( strong measurements ) without using the weak measurement technology @xcite .
in addition , the rapidly varying phase discontinuities along a metasurface , breaking the axial symmetry of the system , enable the direct observation of the spin - dependent splitting @xcite . in the intermediate regime ,
the modified theory is important where neither weak nor strong measurements can detect the spin - dependent splitting .
in conclusion , we have developed a modified weak measurements for the detection of the photonic she . compared with the conventional weak measurements , the amplified shift , amplified factor , and original displacement for different preselected and postselected states
have been examined .
the conventional theory for preselected state @xmath7 is invalid in the strong - coupling regime or when the preselected and postselected states are nearly orthogonal .
but for preselected state @xmath8 it is only limited by the latter since the condition of weak coupling is always satisfied .
we have shown that the weak measurements for detecting photonic she need to be modified when one of the condition is violated .
this is due to the fact that probe wavefunction is distorted in the case of strong coupling or preselected and postselected states are nearly orthogonal .
otherwise , the modified theory can reduce to the conventional one beyond the two restrictions .
we have found that the measuring procedure with preselected and postselected ensembles is still effective .
this scheme is important for us to detect the photonic she in the case where neither weak nor strong measurements can detect the spin - dependent splitting .
our modified theory is valid not only in weak - coupling regime but also in the strong - coupling regime , and especially in the intermediate regime .
we believe that such problem may also exist in the weak measurements of other quantum systems and would have possible applications in precision measurements .
|
weak measurement is an important technique for detecting the tiny spin - dependent splitting in photonic spin hall effect .
the weak measurement is only valid when the probe wavefunction remains almost undisturbed during the procedure of measurements .
however , it does not always satisfy such condition in some practical situations , such as in the strong - coupling regime or the preselected and postselected states are nearly orthogonal . in this paper , we develop a modified weak measurement for detecting photonic spin hall effect when the probe wavefunction is distorted .
we find that the measuring procedure with preselected and postselected ensembles is still effective .
this scheme is important for us to detect the photonic spin hall effect in the case where neither weak nor strong measurements can detect the spin - dependent splitting .
the modified theory is valid not only in weak - coupling regime but also in the strong - coupling regime , and especially in the intermediate regime .
the theoretical models of conventional weak measurements and modified weak measurements are established and compared .
we show that the experimental results coincide well with the predictions of the modified theory .
|
of all the problems quantum gravity is beset with , the most serious is the absence of experimental data even if we had the correct theory how would we know ?
would its beauty alone reveal its truth ? depending on philosophical prejudices sometimes this is answered affirmatively , but more cautious people invoke nature as the ultimate arbitrator . what can be done to satisfy or at least appease the latter in the absence of experiments ?
fortunately , there are some aspects of classical and semi - classical gravity that serve as selection criteria . for instance , a hypothetical theory of quantum gravity failing to reproduce einstein s equations , at least in some limit , would be considered as perverse .
let us now search for some semi - classical selection criterion , because reproducing the einstein equations certainly is laudable but hardly a milestone .
according to @xcite the `` closest thing to experimental data '' we have at our disposal is the bekenstein hawking law , @xmath0 which states that the black hole ( bh ) entropy @xmath1 equals to a quarter of the horizon surface area @xmath2 .
this law has been checked by many independent methods and seems to be so stable against variations in the underlying assumptions that few people would take a putative theory of quantum gravity seriously if it failed to reproduce . in this sense it provides a semi - classical litmus test for any approach to quantum gravity . among the plethora of them
essentially all retrodict the bekenstein hawking law . as carlip pointed out lucidly @xcite : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ `` in a field in which we do not yet know the answers , the existence of competing models may be seen as a sign of health . but the existence of competing models that all _ agree _ cries out for a deeper explanation . ''
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
a possible explanation for the universality of the bekenstein hawking law might stem from near horizon conformal symmetry .
this line of arguments led to numerous ways @xcite of recovering by use of the cardy formula @xcite , which yields a direct relation between entropy and the central charge .
its presence reflects breaking of the conformal symmetry , so the corresponding gauge degrees of freedom should be converted , in a goldstone - like manner , into physical degrees of freedom counted by cardy s formula .
thus , although no particular assumption has been made what the microstates actually are , one is able to establish in a rather model independent way . without a concrete implementation
this is just yet another speculative idea in quantum gravity .
gratifyingly , carlip has found a way to make this program work @xcite in the context of 2d dilaton gravity @xcite . we can not avoid to present the action @xcite , @xmath3\,,\ ] ] but we will need very little of it for the purpose of this essay .
are dyad 1-forms and @xmath4 is essentially the spin - connection .
the free function in , @xmath5 , defines the model and depends on the dilaton @xmath6 and the auxiliary fields @xmath7 .
the schwarzschild bh arises for @xmath8 and @xmath9 ; the witten bh @xcite for @xmath10 and @xmath11 . ] the basic idea now is to ask questions of the form `` if a bh is present , what is the probability of this particular physical process ? '' .
thereby , the bh horizon is implemented as an effective boundary of space - time , which requires specific constraints defining its characteristics .
but as we know since dirac , constraints change the structure of phase space .
carlip could show @xcite that due to the presence of `` stretched horizon constraints '' the algebra of the gauge constraints is changed , as the classical virasoro algebra acquires a central charge proportional to the stretching parameter ( i.e. , it vanishes in the limit of approaching a `` sharp horizon '' ) . in this way
a noteworthy agreement with is found and carlip s attractive proposal appears to work .
but there is something very puzzling about this result : a `` stretched horizon '' does not differ essentially from some generic boundary , so it is not clear in what sense a bh horizon is special .
deriving certainly is commendable , but deriving for an _ arbitrary _ surface `` cries out for some deeper explanation '' . in order to even address this issue
it is mandatory to impose sharp horizon constraints rather than stretched ones .
exactly this task has recently been attacked @xcite and led to a complementary picture . subsequently the surprising result will be recovered and explained , though we impose a simpler variational principle than the gibbons - hawking prescription applied in @xcite .
variation of the action leads not only to the equations of motion , but also to the boundary requirements @xmath12 where @xmath13 denotes the parallel component along the boundary .
this is seen immediately from by tracing all partial integrations .
generically @xmath14 is the only choice possible , but there exists a crucial exception : one can prove that a killing horizon emerges if either @xmath15 or @xmath16 vanish @xcite , regardless of the gauge .
thus , a consistent set of sharp horizon boundary conditions is given by @xmath17 it is remarkable that horizons differ from generic boundaries already at the level of the variational principle .
the cases and are implemented via boundary constraints besides the primary and secondary bulk constraints generating gauge symmetries .
quite generically , this effects the bulk constraints to become second class .
however , it is found that a horizon entails several first class constraints and consequently gauge symmetries .
a study of the latter reveals diffeomorphisms along the boundary and local lorentz transformations to be unrestricted , concurrent with our intuition about null hyper - surfaces .
this provides another hint that sharp horizons behave profoundly different from generic boundaries .
still one should be careful as the relevant contributions to the poisson brackets between the constraints have support only at the boundary .
therefore naive counting arguments can not be invoked to get the respective numbers of physical degrees of freedom .
instead we have to construct the reduced phase space explicitly as described detailed in @xcite .
it is illuminating to pinpoint the pivotal technical observations : during the construction most boundary constraints are found to be obsolete as they simply represent the analytic continuation of the bulk gauge fixing to the boundary . in this way _ all _ boundary constraints disappear for a generic boundary , leaving two free functions which comprise the ( physical ) boundary phase space , in accordance with a well - known result by kucha @xcite .
they can be interpreted as mass and its conjugate , the `` proper time '' at the boundary .
indeed , all 2d dilaton gravity models exhibit a constant of motion @xcite directly related to the adm mass . in order to reduce its boundary value to the core part ,
@xmath18 the trivial shift and scaling ambiguities inherent to any mass definition have been fixed by choosing a convenient ground state and energy scale , respectively .
from there is no restriction on @xmath19 at the boundary , so @xmath20 may fluctuate off - shell .
however , the sharp horizon boundary conditions uniquely determine the constraint @xmath21 which is first class and thus represents a boundary gauge degree of freedom .
the `` proper time '' is pure gauge as well because of residual lorentz transformations .
consequently the physical boundary phase space is empty for horizons .
the conclusion is inevitable : the physical degrees of freedom present on a generic boundary are converted into gauge degrees of freedom on a horizon .
at first glance our results seem to contradict earlier findings , as apparent already from the respective section titles . in order to reconcile them
it may be tempting to interpret the disaccord as a manifestation of bh complementarity @xcite : for the outside observer the horizon is not accessible anyhow and thus a stretched horizon could provide an adequate description consistent with simple thermodynamical considerations .
however , by the same token one could argue that there should be something wrong with `` sharp '' asymptotic conditions because no physical observer can get access to the infinite far region .
evidently a more convincing interpretation is needed . since the reduced phase space is non - empty for _ any _ generic boundary it seems unlikely to us that the corresponding physical degrees of freedom bear a _
direct _ relation to bh entropy . as we have shown this information
is lost on a horizon by transmutation due to enhanced symmetry . because the very notion of entropy is built upon ( non - accessible ) information it is plausible at a qualitative level why emerges . to obtain it
quantitatively one can still employ the cardy formula and proceed e.g. by analogy to @xcite , but our results put that procedure in a quite different perspective .
though entropy appears to stem from goldstone modes due to breaking of conformal symmetry , our analysis demonstrates that they actually mimic physical modes which were there in the first place , but which have been converted on a horizon into something unobservable , namely gauge degrees of freedom .
this resonates strikingly with a proposal by t hooft @xcite : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ `` a more subtle suggestion is that , although we do have fields between the [ brick ] wall and the horizon , which do carry degrees of freedom , these degrees of freedom are not physical .
they could emerge as a kind of _ local gauge degrees of freedom _ ,
undetectable by any observer . ''
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ it is reassuring that we could establish the validity of this conjecture for bhs permitting an effective 2d description as in , including schwarzschild .
since also more complicated scenarios allow such a treatment , at least in the vicinity of the horizon , the features discussed in this essay ought to be universal for bhs . to summarize our findings in a single sentence : _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ entropy arises because approaching the black hole horizon does not commute with constructing the physical phase space . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
we thank cordially wolfgang kummer and dima vassilevich for collaboration on related topics , in particular on @xcite .
dg has been supported by project gr-3157/1 - 1 of the german research foundation ( dfg ) and during the final preparations of this paper also by the marie curie fellowship mc - oif 021421 of the european commission under the sixth eu framework programme for research and technological development ( fp6 ) .
s. carlip , `` horizons , constraints , and black hole entropy , '' http://www.arxiv.org/abs/gr-qc/0601041[gr-qc/0601041 ] .
a. strominger , _ jhep _ * 02 * ( 1998 ) 009 , http://www.arxiv.org/abs/hep-th/9712251[hep-th/9712251 ] .
d. birmingham , i. sachs , and s. sen , _ phys .
lett . _ * b424 * ( 1998 ) 275280 , http://www.arxiv.org/abs/hep-th/9801019[hep-th/9801019 ] .
see also @xcite and references therein .
j. l. cardy , _ nucl .
_ * b270 * ( 1986 ) 186204 .
s. carlip , _ class .
* 22 * ( 2005 ) 13031312 , http://www.arxiv.org/abs/hep-th/0408123[hep-th/0408123 ] .
d. grumiller , w. kummer , and d. v. vassilevich , _ phys .
* 369 * ( 2002 ) 327429 , http://arxiv.org/abs/hep-th/0204253[hep-th/0204253 ] .
schaller and t. strobl , _ mod .
lett . _ * a9 * ( 1994 ) 31293136 .
http://arxiv.org/abs/hep-th/9405110[hep-th/9405110 ] .
e. witten , _ phys .
rev . _ * d44 * ( 1991 ) 314324 .
cf . c. g. callan , jr .
, s. b. giddings , j. a. harvey , and a. strominger , _ phys . rev . _
* d45 * ( 1992 ) 10051009 , http://www.arxiv.org/abs/hep-th/9111056[hep-th/9111056 ] .
l. bergamin , d. grumiller , w. kummer , and d. v. vassilevich , _ class .
* 23 * ( 2006 ) 30753101 , http://www.arxiv.org/abs/hep-th/0512230[hep-th/0512230 ] .
k. v. kucha , _ phys .
rev . _ * d50 * ( 1994 ) 39613981 , http://www.arxiv.org/abs/gr-qc/9403003[gr-qc/9403003 ] .
g. t hooft , `` horizons , '' http://www.arxiv.org/abs/gr-qc/0401027[gr-qc/0401027 ] .
|
frequently it is argued that the microstates responsible for the bekenstein hawking entropy should arise from some physical degrees of freedom located near or on the black hole horizon . in this essay
we elucidate that instead entropy may emerge from the conversion of physical degrees of freedom , attached to a generic boundary , into unobservable gauge degrees of freedom attached to the horizon . by constructing the reduced phase space
it can be demonstrated that such a transmutation indeed takes place for a large class of black holes , including schwarzschild .
|
statistical deposition models have attracted much attention in the last years because they may describe real systems features by representing the basic growth mechanisms as simple stochastic processes , thus neglecting the details of the microscopic interactions @xcite .
some examples are the family model @xcite , the restricted solid - on - solid ( @xmath4 ) model of kim and kosterlitz @xcite and the ballistic deposition ( @xmath9 ) model @xcite .
the former one is representative of the edwards - wilkinson ( @xmath10 ) universality class @xcite of linear growth , while the other ones are representative of the kardar - parisi - zhang ( @xmath11 ) class @xcite of non - linear growth .
more complex models involve the competition among deposition , diffusion and aggregation @xcite and may represent quantitatively real systems features .
often one is interested in surface properties , then the main quantity to be measured is the interface width of the deposit . if deposition occurs in a @xmath12-dimensional substrate of length @xmath13 , then the interface width at time @xmath14 is defined as @xmath15 } ^{1/2 } } \right > } \label{eq:1}\ ] ] or as @xmath16 } ^{1/2 } .
\label{eq:2}\ ] ] in eqs .
( 1 ) and ( 2 ) , @xmath17 is the height of column @xmath18 , the bar in @xmath19 denotes a spatial average and the angular brackets denote a configurational average , i. e. , an average over many realizations of the noise . @xmath20 and
@xmath21 have different values but the same scaling properties , then in the following any definition or discussion concerning @xmath20 is also valid for @xmath21 . at early times , in the so - called growth regime ,
the interface width increases as @xmath22 while at long times finite - size effects lead to its saturation at @xmath23 relations ( 3 ) and ( 4 ) are included in the dynamic scaling relation proposed by family and vicsek @xcite , @xmath24 in which @xmath25 is a scaling function and the dynamical exponent @xmath2 is given by @xmath26 in eq .
( 5 ) , the argument of the scaling function is a ratio of the deposition time and the crossover ( or saturation ) time @xmath27 .
most numerical works on deposition models focus on the calculation of exponents @xmath28 ( eq .
4 ) and @xmath29 ( eq .
3 ) - see e. g. the summary of numerical results in ch .
the calculation of @xmath28 requires the identification of the steady - state regime , in which @xmath20 saturates . eventually , the extrapolation of effective exponents is essential , such as in the @xmath9 model @xcite .
the calculation of @xmath29 depends on the definition of the growth regime , which must be a region of high linear correlation in a @xmath30 plot .
finding such region may be very difficult for @xmath31 due to small lattice lengths @xcite .
the usual methods to estimate exponent @xmath2 are based on data collapse of the scaled quantities appearing in eq .
( 5 ) or on the scaling of the structure factor @xmath32 ( @xmath33 is the fourier transformation of the surface height @xmath34 ) @xcite . in a recent work ,
kim et al presented a method to estimate @xmath2 from the exponential decay of a relaxation function , starting the growth process from a sinusoidal surface @xcite . in @xmath35
, they obtained estimates for the family , the @xmath4 and the krug models which were consistent with the expected universality classes @xcite .
however , for the most common problem of deposition in a flat surface , a precise definition of the crossover time @xmath0 and a method to estimate @xmath2 accurately are still lacking .
it is also important to recall that there are controversies on the universality classes of various statistical growth models even in @xmath35 @xcite . at this point ,
the development of reliable techniques to estimate critical exponents is also helpful .
the aim of this work is to analyze the convergence of the scaling function in eq .
( 5 ) to an asymptotic constant value ( consistently with eq .
4 ) and to propose and test methods to estimate crossover times and the exponent @xmath2 . first we will consider the family , the @xmath4 and the @xmath9 models in @xmath35 .
we will show that the typical decay of @xmath20 to its saturation value is exponential , which uniquely defines the crossover or saturation time @xmath0 .
then we will present a method to calculate a characteristic time @xmath1 that is proportional to @xmath0 , and we will estimate the exponent @xmath2 from @xmath1 . the reliability of the method will be proved with comparisons with exact results for the corresponding hydrodynamic equations .
we will also consider the family and the @xmath4 models in @xmath36 .
results for the @xmath4 model confirm the exponential decay of @xmath20 to the saturation value and give @xmath7 . estimates of the exponent @xmath28 , obtained from @xmath37 data , are in good agreement with the prediction of marinari et al @xcite that @xmath38 for @xmath11 in @xmath36 .
for the family model , we will consider the expected dynamic scaling hypothesis , which involves a @xmath39 growth , and we will show results consistent with @xmath6 .
a recent numerical work @xcite showed large discrepancies from that value , using data collapse techniques , which proves the relevance of the methods presented here .
the rest of this paper is organized as follows . in sec
ii we present the results of simulations in @xmath35 for the family , the @xmath4 and the @xmath9 models , show the exponential decay of @xmath20 to its saturation value and the methods to calculate @xmath2 . in sec .
iii we present the simulations results of the family and the @xmath4 models in @xmath36 , including estimates of exponent @xmath28 for the @xmath4 model . in sec .
iv we summarize our results and present our conclusions .
first we recall the definition of the models studied here . in the family model @xcite , a column of the deposit is randomly chosen and the incident particle searches for the smallest height in its neighborhood . if no neighboring column has a smaller height than the column of incidence , the particle sticks at the top of this one .
otherwise , it sticks at the top of the column with the smallest height among the neighbors .
if two or more neighbors have the same minimum height , the sticking position is randomly chosen among them . in the @xmath4 model
@xcite , the incident particle may stick at the top of the column of incidence if the differences of heights of all pairs of neighboring columns do not exceed @xmath40 .
otherwise , the aggregation attempt is rejected . in @xmath9 ,
the incident particle follows a straight trajectory perpendicular to the surface and sticks upon first contact with a nearest neighbor occupied site .
it leads to the formation of a porous deposit , contrary to the previous solid - on - solid models . in all models
, a time unit corresponds to the deposition of @xmath13 particles .
we simulated them in one - dimensional substrates with lengths @xmath41 , with integer @xmath42 ( @xmath43 ) .
for each @xmath13 , typically @xmath44 different deposits were generated , and very long steady - state regions were observed . in order to obtain reliable estimates of @xmath37 , we computed it over different time ranges to ensure that it was fluctuating around an average value , and not increasing systematically in time ( the latter possibility would suggest that the steady - state region was not attained yet ) .
@xmath37 was obtained with accuracy smaller than @xmath45 in all cases , which enables the calculation of the deviation from @xmath37 , @xmath46 also with good accuracy . in fig .
1a we show @xmath47 for the family model in @xmath48 , and in fig .
1b we show @xmath47 for the @xmath4 and @xmath9 models , also in @xmath48 . in figs .
2a and 2b we show @xmath49 in the same systems .
it clearly shows an exponential decay of @xmath20 and @xmath21 to @xmath37 and @xmath50 , in the form @xmath51 with constants @xmath52 and @xmath0 .
the estimates of relaxation times shown in figs . 1 and 2 were obtained from fits of the data in regions of high linear correlation . for the other lengths @xmath13 ,
similar linear behaviors in @xmath47 plots were obtained .
the calculation of @xmath0 from those plots has some disadvantages .
for instance , @xmath0 has to be estimated in different time ranges , in order to avoid spurious effects of an arbitrary choice of the fitting region .
thus we searched for an alternative method , as described below , which is not based on similar data fits . the scaling relation ( 5 ) and eq .
( 8) imply that the amplitude @xmath52 of eq .
( 8) is proportional to @xmath53 .
consequently , when approaching the steady - state region , the scaling function of eq .
( 5 ) may be written as @xmath54 with constant @xmath55 and @xmath56 . if a characteristic time @xmath1 is defined from @xmath57 with constant @xmath58 , then @xmath59 } , \label{eq:11}\ ] ] i. e.
, @xmath1 is proportional to the crossover time @xmath0 .
the estimates of @xmath1 from eq .
( 10 ) always had higher accuracy than the relaxation times obtained from @xmath60 plots ( typically for a factor @xmath61 ) .
we considered two possible values of the constant @xmath58 in eq .
( 10 ) , @xmath62 and @xmath63
. the first choice would give @xmath64 if @xmath65 ( eqs .
9 and 11 ) , which seems to be approximately valid for @xmath9 .
the effective exponents @xmath5 were obtained from @xmath66 \over \ln 2 } , \label{eq:12}\ ] ] and must be independent of the particular choice of the constant @xmath58 . in figs .
3a and 3b we show @xmath5 versus @xmath67 for the three models using @xmath68 and @xmath69 , respectively , in the calculation of @xmath1 ( eq .
@xmath20 data were used , but the effective exponents obtained from @xmath21 are nearly the same .
the uncertainties in @xmath37 are responsible for the error bars , thus results with @xmath70 ( larger @xmath71 ) are more accurate . in all cases shown in figs .
3a and 3b , @xmath5 converges to the exactly known values as @xmath72 . in the family and in the @xmath4 models , @xmath6 @xcite and @xmath73 @xcite
are recovered with small finite - size corrections . in @xmath9 ,
@xmath67 corrections to @xmath5 lead to expected @xmath73 @xcite asymptotic value .
it is interesting to notice that these corrections in @xmath9 are weaker than the corrections found in effective exponents @xmath74 and @xmath75 in ref .
@xcite , which proves the relevance of developing methods to estimate @xmath2 independently .
moreover , this result complements that systematic study of @xmath9 in @xmath35 , which discussed the discrepancies in numerical estimates of critical exponents and the controversies on the equivalence of @xmath9 and the @xmath11 theory @xcite .
first we consider the @xmath4 model in @xmath36 .
simulations were performed for lengths @xmath76 , with a small number of realizations ( nearly @xmath77 ) in the largest lattices .
the exponential decay of @xmath78 is confirmed in fig .
4a for lattices of length @xmath79 . in fig .
4b we show @xmath5 versus @xmath67 , which gives the asymptotic estimate @xmath80 .
this result is consistent with several previous estimates of critical exponents for that model @xcite . using @xmath37 data , we were also able to obtain finite - size estimates of the exponent @xmath28 : @xmath81 \over \ln 2 } . \label{eq:13}\ ] ] in figs . 5a and
5b we show @xmath75 versus @xmath67 obtained from @xmath20 and @xmath21 data , respectively .
both sets of data strongly suggest an asymptotic exponent @xmath8 , which agrees with simulations of marinari et al @xcite , who obtained @xmath82 .
however , it disagrees with the proposal of @xmath83 based on an operator product expansion @xcite , which suggests further investigations on these lines . for the family model , it is expected the dynamic scaling relation @xcite @xmath84 } \label{eq:14}\ ] ] where @xmath25 is a scaling function and @xmath52 is constant . as @xmath85 , it is expected that @xmath86 , with constant @xmath55 . the characteristic time @xmath1 defined from eq .
( 10 ) is not proportional to @xmath0 anymore .
then , in order to define a characteristic time @xmath87 which is proportional to @xmath0 in this system , we consider @xmath88 with constant @xmath89
. the characteristic times @xmath87 were calculated using eq .
( 15 ) with @xmath90 .
the effective exponents @xmath5 were also obtained from eq .
( 12 ) . in figs .
6a and 6b we show @xmath5 versus @xmath67 obtained from @xmath20 and @xmath21 data , respectively . in both cases
, @xmath5 converges to @xmath6 as @xmath72 , in agreement with the @xmath10 theory .
our result is significantly better than the value @xmath91 , obtained in ref .
@xcite using data collapse methods , which again proves the reliability of our method .
we proposed and tested methods to estimate crossover times of interface width scaling and the dynamic exponent @xmath2 .
we showed that the typical decays of @xmath20 or @xmath21 to their saturation values are exponential , which uniquely defines the crossover or saturation times @xmath0 . in order to estimate @xmath2 with accuracy
, we used a characteristic time @xmath1 that is proportional to @xmath0 , according to the family - viczek scaling .
the reliability of the method was proved by testing it with the family , the @xmath4 and @xmath9 models in @xmath92 dimensions , in which dynamic exponents of the @xmath10 and the @xmath11 theory were obtained . the family and the @xmath4 models in @xmath36 were also considered . for the @xmath4 model , the exponential decay of @xmath20
is confirmed , and @xmath7 was obtained .
moreover , estimates of saturation widths suggest that @xmath8 , in agreement with a recent prediction of large scale simulations @xcite . for the two - dimensional family model ,
the expected dynamic scaling hypothesis lead to an alternative definition of the characteristic time @xmath1 , and the exact value @xmath6 was recovered .
the relevance of our methods to estimate dynamical exponents is also proved by comparisons with previous results for related systems .
for instance , using data collapse of relaxation functions and the expected scaling of @xmath20 at @xmath93 , @xmath94 was obtained for eden clusters @xcite and @xmath95 for the single - step model @xcite in @xmath35 .
both systems are in the @xmath11 class ( exact @xmath96 ) .
these difficulties were pointed out in the paper of kim et al @xcite , who proposed a method to measure @xmath2 from the scaling of a relaxation function and an initial sinusoidal surface . although the efficiency of their method was clearly stated , working with a flat surface is the most usual initial situation in theoretical works and is more suitable for real applications . moreover , with our method , one is able to calculate three exponents , @xmath28 , @xmath2 and @xmath29 , using the same set of data .
the relevance of each one to determine the universality class of a particular model may depend on several factors , such as the strength of finite - size corrections .
thus we expect that this work may be useful for further studies of surface growth models .
this work was partially supported by cnpq and finep ( brazilian agencies ) .
the author thanks the department of theoretical physics at oxford university , where part of this work was done , for the hospitality . a .-
barabsi and h. e. stanley , fractal concepts in surface growth ( cambridge university press , new york , 1995 ) .
f. family , j. phys . a * 19 * ( 1996 ) l441 . s. f. edwards and d. r. wilkinson , proc .
london * 381 * ( 1982 ) 17 . j. m. kim and j. m. kosterlitz , phys .
* 62 * ( 1989 ) 2289 .
m. j. vold , j. coll .
* 14 * ( 1959 ) 168 ; j. phys . chem . * 63 * ( 1959 ) 1608 .
f. family and t. vicsek , j. phys . a * 18 * ( 1985 ) l75 . m. kardar , g. parisi and y .- c .
zhang , phys .
* 56 * ( 1986 ) 889 .
m. c. bartelt and j. w. evans , phys .
b * 46 * ( 1992 ) 12675 . s. clarke and d. d. vvedensky , j. appl . phys .
* 63 * ( 1988 ) 2272 .
c. ratch , a. zangwill , p. smilauer , and d. d. vvedensky , phys .
* 72 * ( 1994 ) 3194 .
f. d. a. aaro reis , phys .
e * 63 * ( 2001 ) art .
no . 056116 .
m. plischke and z. rcz , phys .
a * 32 * ( 1985 ) 3825 .
m. plischke , z. rcz , and d. liu , phys .
a * 35 * ( 1987 ) 3485 .
t. ala - nissila , t. hjelt , j. m. kosterlitz , and o. venlinen , j. stat . phys .
* 72 * ( 1993 ) 207 .
j. m. kim , j. h. lee , i. kim , j. yang , and y. lee , physica a * 278 * ( 2000 ) 304 .
r. a. blythe and m. r. evans , phys .
e * 64 * ( 2001 ) art .
no . 051101 .
e. marinari , a. pagnani and g. parisi , j. phys .
a * 33 * ( 2000 ) 8181 .
s. pal and d. p. landau , physica a * 267 * ( 1999 ) 406
. m. lssig , phys .
* 80 * ( 1998 ) 2366 .
b. m. forrest and l .- h .
tang , j. stat . phys . * 60 * ( 1990 ) 181 .
|
considering some deposition models with limited mobility , we show that the typical decay of the interface width to its saturation value is exponential , which defines the crossover or saturation time @xmath0 .
we present a method to calculate a characteristic time @xmath1 proportional to @xmath0 and estimate the dynamical exponent @xmath2 . in one dimensional substrates of lengths @xmath3 , the method is applied to the family model , the restricted solid - on - solid ( @xmath4 ) model and the ballistic deposition .
effective exponents @xmath5 converge to asymptotic values consistent with the corresponding continuum theories .
for the two - dimensional family model , the expected dynamic scaling hypothesis suggests a particular definition of @xmath1 that leads to @xmath6 , improving previous calculations based on data collapse methods .
for the two - dimensional @xmath4 model , we obtain @xmath7 and @xmath8 , in agreement with recent large scale simulations .
|
the binary millisecond pulsar psr j1740 - 5340 was recently discovered in the globular cluster ngc 6397 by damico et al .
( 2001a ) with the parkes radio telescope . they reported a spin period of @xmath5ms , a @xmath6 day circular orbit , and eclipses at @xmath7 ghz that last for more than 40% of the orbital period .
subsequent parkes observations ( damico et al .
2001b ) allowed an accurate timing solution yielding the spin - down rate as well as the binary s coordinates . with the position known , ferraro et al .
( 2001 ) used archived images to obtain a secure optical identification with a star ( wf4 - 1 ) previously listed as a by dra candidate ( taylor et al .
the pulsar s companion ( named com j1740 - 5340 by ferraro et al .
2003 ) has a luminosity close to that of the cluster s main - sequence turnoff stars but its colour is anomalously red . in addition , the companion is variable with a range of @xmath8mag .
this variability has the same period as the binary orbit and matches that due to the tidal distortion of a star filling its roche lobe .
ferraro et al . ( 2001 ) emphasize that , in this regard , psr j1740 - 5340 differs from two other eclipsing millisecond pulsars , where strong photometric variability is clearly due to the heating of the companions facing hemispheres by the pulsars emission . the absence of this expected heating effect is also reported by kaluzny et al .
( 2003 ) and by orosz & van kerkwijk ( 2003 ) .
although not apparent in broad - band light curves , the effect of the pulsar s emission on its companion has almost certainly been detected spectroscopically by ferraro et al .
( 2003 ) . from observations made to determine the spectroscopic orbit
, these authors made the serendipitous discovery of he@xmath0i absorption lines at 5876 and 6678 .
given that the companion s effective temperature t@xmath9k ( ferraro et al . 2003 ) is substantially less than the t@xmath10 ( @xmath11k ) at which the he@xmath0i lines are present for main sequence stars , this discovery is a major surprise and obviously significant .
ferraro et al . ( 2003 ) explain the presence of he@xmath0i absorption in terms of a narrow strip across the illuminated face of the companion within which the local t@xmath12k due to heating by beamed pulsar radiation . but outside this strip , irradiation is negligible and so the local t@xmath10 is then @xmath13 . by a suitable choice of parameters ,
ferraro et al . imply that the observed strengths of the @xmath15876line can reconciled with the absence of a photometric heating effect . in a very recent paper , this group ( sabbi et al . 2003 ) have published more of their spectroscopy and repeated their conclusion that the pulsar s emission pattern must be highly anisotropic .
although the model proposed by ferraro et al . is not implausible , another effect suggests itself if we note that com j1740 - 5340 has two points in common with type ib sne .
the first is that the spectra of these sne are also remarkable in exhibiting he@xmath0i absorption lines despite cool photospheric continua ( harkness et al .
1987 ) .
the second point in common - the presence of high - energy photons - follows from the explanation for the helium lines in the ib s . in the standard interpretation ( lucy 1991 , hereafter paper i ; swartz et al .
1993 ) , a huge overpopulation of the @xmath14 levels is caused by impact excitations and ionizations of helium atoms by non - thermal electrons .
these energetic electrons are the cascade products created by the multiple compton scatterings of @xmath2-rays emitted by the radioactive nuclei @xmath15co .
there is a significant energy density of such @xmath2-rays in the reversing layers of these sne because of mass loss preceding the explosion and the mixing out of @xmath15ni following the explosion . in view of these similarities ,
an obvious conjecture is that @xmath2-rays emitted by the magnetosphere of psr j1740 - 5340 , or arising when the pulsar s relativistic wind encounters the companion s wind , irradiate the atmosphere of its companion , overpopulate the @xmath14 levels of he@xmath16 , and hence give rise to detectable absorption lines .
the investigation of this conjecture is the subject of this paper .
in this section , the problem is defined and then the steps needed to calculate the he line spectrum are briefly described .
the discussion follows closely that given in paper i for type ib sne . in investigating the non - thermal excitation of helium in the atmosphere of com j1740 - 5340 ,
we face several major uncertainties .
first and foremost , there is no direct evidence that psr j1740 - 5340 emits a powerful beam of hard radiation . although the object is detected in x - rays ( grindlay et al . 2001 , 2002 ) , its x - ray luminosity @xmath17 is weaker by a factor @xmath18 than expected for its spin - down luminosity @xmath19 from the correlation @xmath20 found for millisecond pulsars in the field ( becker & trmper 1997 ) .
however , the detected x - rays are probably not emitted from the pulsar s magnetosphere but rather from the interaction region between the pulsar s relativistic wind and the mass loss from the companion ( grindlay et al .
2002 ) . despite this absence of evidence for a beam of hard radiation
, we assume that such a beam exists and that it sweeps over at least part of the companion s disk .
accordingly , the problem investigated in this paper is to study the formation of the he@xmath0i spectrum at a typical _ point _ on the companion s disk lying within the strip swept out by the beam . the problem as posed refers to irradiation by a pencil beam emitted by the pulsar s magnetosphere .
but the calculation is also relevant to irradiation by an extended source of hard radiation , such as that due to the colliding winds of the pulsar and its companion .
the single beam then represents a one - point quadrature approximation for the effect of the extended source . because everything about the beam is conjectural ,
we take it to be mono - energetic , comprising photons of single energy @xmath21 , the beam s effective photon energy .
if the beam s flux as a function @xmath22 were known , @xmath21 could be defined as the value of @xmath22 such that the mono - energetic beam deposits the same fraction of its flux in the reversing layer as would the actual beam .
nevertheless , since the vertical distribution of the deposited energy would not be exact , the concept of an effective photon energy is not rigorous . the phase - averaged flux incident on the companion s atmosphere is conveniently written as @xmath23 where @xmath24 is the incident direction cosine
, @xmath25 is the orbital separation , @xmath26 is the conversion efficiency of @xmath19 into beamed radiation , and @xmath27 is the solid angle swept out by the beam .
the adopted numerical values are as follows : @xmath28 erg s@xmath29 ( damico et al .
2001b ) ; @xmath30 cm , corresponding to component masses @xmath31 = 1.5 @xmath32 , @xmath33 = 0.26 @xmath32 ( ferraro et al . 2003 ) ; and @xmath34 steradian .
this last choice is standard practice in view of our ignorance concerning beaming fractions . inserting these values , we have @xmath35 erg @xmath36 s@xmath29
. a full nlte investigation of the response of the companion s atmosphere to irradiation by a beam of @xmath2 rays is not justified in view of the above - mentioned uncertainties .
accordingly , the approach adopted is to embed a nlte calculation for he atoms in an atmosphere that is otherwise unperturbed by irradiation .
moreover , a severely simplified model is adopted for this unperturbed atmosphere .
specifically , a plane - parallel , schuster - schwarzschild model is assumed comprising a black - body emitting photosphere at rosseland mean optical depth @xmath37 and an isothermal reversing layer with temperature @xmath38 . on the assumption of hydrostatic equilibrium and
constant rosseland mean opacity @xmath39 , the variation of gas pressure in the reversing layer is given by @xmath40 , where @xmath41 is the surface gravity .
the adopted constant opacity @xmath39 is the solution of the equation @xmath42 .
the parameters of the model atmosphere are @xmath43 ( ferraro et al .
2003 ) , @xmath44 , x=0.75 and y=0.25 .
the value of @xmath45 corresponds to radius @xmath46 ( orosz & van kerkwijk 2003 ) and mass @xmath33 = 0.26 @xmath32 .
the adopted chemical composition neglects metals since [ fe / h]@xmath47 ( ferraro et al . ) and assumes that envelope stripping has not yet exposed he - enriched zones . with these parameters ,
the equation for the opacity has the solution @xmath48 dex .
this was obtained by iterative interpolation in table 6 of alexander & ferguson ( 1994 ) .
although this table is computed for @xmath49 , the h@xmath50 ion is the dominant source of opacity for the given photospheric parameters , with metals as minor contributors to the electron density @xmath51 . in the subsequent calculations , the reversing layer is discretized into 50 layers of width @xmath52 , within each of which physical variables are constant .
the resulting small optical depths of the outermost layers ensure that line centres are accurately calculated .
the first step in calculating the he spectrum is to derive the rate at which the beam deposits energy in the reversing layer in the form of energetic electrons .
each incident @xmath2-ray creates a cascade of non - thermal electrons by multiple compton scatterings off free and bound electrons . at optical depth
@xmath53 , the rate at which @xmath2-rays are losing energy by compton scattering is @xmath54 where @xmath55 is the compton scattering coefficient , @xmath56 is the expected fraction of @xmath22 transferred to the compton electron , and @xmath57 is the mean intensity at photon energy @xmath22 .
note that here , and throughout the paper , rates refer to unit volume . as in paper i
, a monte carlo code is used to simulate this process .
if the incident flux @xmath58 at @xmath59 is represented by @xmath60 identical photon packets containing @xmath2-rays of energy @xmath21 and entering the atmosphere in time interval @xmath61 , then each packet s initial energy @xmath62 is given by @xmath63 .
these packets are followed as they propagate within the reversing layer until they escape at @xmath59 or penetrate below the photosphere at @xmath37 , where they are assumed to thermalize .
when the scattering histories of all @xmath60 packets are complete , the loss rate in the @xmath64-th layer is derived with the monte carlo estimator ( lucy 1999 , 2003 ) @xmath65 here @xmath66 is the volume of a vertical column of unit cross section through the @xmath64-th layer , and @xmath67 is the pathlength of a packet of energy @xmath68 between consecutive events , where an event is either a compton scattering or a crossing of a layer s boundary .
the summation is over all pathlengths in layer @xmath64 .
monte carlo convergence experiments show that this estimator gives fractional errors @xmath69 for @xmath70 , the sample size adopted here .
note that this estimator returns a non - zero value for the loss rate even if no @xmath2-ray packet undergoes a compton scattering in layer @xmath64 . because @xmath71 is proportional to @xmath26 , it is convenient to compute its value @xmath72 for @xmath73 .
the energy deposition rate in layer @xmath64 is then @xmath74 .
the next step is to determine the rates of impact excitations and ionizations of he atoms . to achieve this
, we must follow the slowing down of fast electrons as they interact with matter in the reversing layer and record the energy transfers into the various deposition channels . this calculation
is greatly facilitated by two approximitions adopted in paper i. first , noting that the stopping distance for an mev electron is @xmath75 that for an mev photon ( colgate , petschek & kriese 1980 ) , we assume that the electrons deposit all their energy in situ .
the second approximation ( meyerott 1978 ) makes use of the asymptotic near constancy of @xmath76 , the fraction of an electron s kinetic energy @xmath77 that is deposited in channel @xmath78 .
thus we need only compute the deposition fractions @xmath79 for @xmath80kev since @xmath81 for @xmath82 .
the values @xmath79 are obtained by solving the spencer - fano equation - eq ( 2 ) in paper i. this requires that we identify the stopping mechanisms and provide their cross sections . in paper
i , this equation was solved for a weakly - ionized pure helium gas , so that the stopping mechanisms were impact excitations and ionizations of he@xmath16 atoms and coulomb scattering with free ( thermal ) electrons .
now the equation must be solved for a weakly - ionized h - he gas , so the stopping mechanisms to be added are impact excitations and ionizations of h atoms .
the treatment of he impacts and coulomb scattering remain as described in sect .
2.3 of paper i. the added impact cross sections for h are the approximate formulae of johnson ( 1972 ) , with the h atom represented by 14 bound levels with principal quantum numbers @xmath83 . in addition , the probability distribution for the energy @xmath84 of the secondary electron when h is ionized is taken to be @xmath85 with @xmath86ev ( shull 1979 ) . with composition specified ,
the deposition fractions @xmath79 are functions of just one variable @xmath87 , the fractional ionization of h. in conformity with the assumption ( sect .
2.3 ) that the atmosphere is unperturbed by irradiation , electron densities are computed from saha s equation . then , neglecting the variation with height , we set @xmath88 , its value at @xmath89 . with this approximation ,
the deposition of non - thermal energy throughout the reversing layer is governed by the single vector @xmath90 .
summing the appropriate elements of this vector , we find that 11.7% of the deposited energy is converted directly into heat via coulomb scattering , 82.0% is converted into ionization and excitation energy of h , and the remaining 6.3% is converted into ionization and excitation energy of he .
the rate at which non - thermal energy is absorbed in exciting he atoms to level @xmath91 in layer @xmath64 is @xmath92 .
accordingly , if @xmath93 is the energy of this level relative to the ground state , the rate of impact excitation ( ionization ) of level @xmath91 in layer @xmath64 is @xmath94 if @xmath95 denotes the rate at which transitions @xmath96 occur due to conventional radiative and collisional ( thermal ) processes , then , with impacts included , the equations of statistical equilibrium for he levels @xmath97 are @xmath98 when the constraint @xmath99 is added , we have @xmath100 equations in the @xmath100 unknowns @xmath101 . this system must be solved in each layer of the atmosphere in order to compute the emergent profiles of the he lines .
but since the radiative coefficients in eq .
( 5 ) depend on the solutions of the line transfer problems , iterations are required ( sect .
the data sources for the @xmath102 coefficients in eq .
( 5 ) are as stated in paper i , except that the collision strengths of transitions between the lowest five levels of he have been updated using the calculations of sawey & berrington ( 1993 ) . for most of the
he lines , we can assume that line photons in the reversing layer interact only with their own transition - i.e. , continuum process can be neglected .
but this is not true for transitions to the ground state since the emitted photons can ionize hydrogen and the atmosphere is optically thick in the lyman continuum . for the adopted atomic model ,
this is relevant for the permitted ground state transitions @xmath4 at 584 , 537 and 522 .
since the optical depths for these transitions are also huge , we can assume that when such a line photon is emitted it is absorbed in situ either by the line itself or by a h atom .
accordingly , we adopt an escape probability formalism for these three transitions and only consider thermal motions in computing doppler broadening .
if @xmath103 denotes the normalized absorption profile , a line photon emitted with frequency @xmath104 sees a line absorption coefficient @xmath105 , where @xmath106 and a continuum absorption coefficient @xmath107 , where we assume all h atoms are neutral and in the ground state and neglect the correction for stimulated emission .
the probability that this particular photon will be lost to the line is therefore @xmath108 .
thus , if we average over the emission profile , the escape probability is @xmath109 note that the emission profile is taken to be identical to the absorption profile on the assumption of complete redistribution ( mihalas 1978 , p.29 ) .
this treatment is applied to the permitted transitions @xmath110 for @xmath111 . as a result , the radiative contribution of these transitions to eq .
( 5 ) simplifies to @xmath112 , where @xmath113 is the einstein coefficient for spontaneous emission . for the isothermally stratified reversing layer ,
these escape probabilities are independent of @xmath53 . with @xmath103 given by the voigt function assuming pure radiation damping
, we obtain @xmath114 = @xmath115 and @xmath116 for the lines at 584 , 537 and 522 , respectively . an escape probability treatment is also adopted for recombinations to the ground state .
the emitted photon will travel a negligible distance before it photoionizes a h or a he atom .
since the recombination photons are strongly concentrated to the he ionization threshold at @xmath117 , the escape probability is accurately given by ( osterbrock 1974 , p.23 ) @xmath118 accordingly , if @xmath119 denotes the total recombination coefficient for the ground state , it is replaced in the equations of statistical equilibrium by @xmath120 . with the he threshold cross section @xmath121@xmath122 of fernley et al .
( 1987 ) , we obtain @xmath123 subordinate transitions are not treated with escape probabilities but by solving the transfer equation . for such transitions , the radiative contribution to eq .
( 5 ) is @xmath124 , where @xmath125 is the profile - averaged mean intensity .
this quantity is obtained by numerical integration after solving the line transfer equation @xmath126 with source function @xmath127 for a grid of frequencies spanning the transition . since high precision is not required , eq .
( 10 ) is solved only at @xmath128 giving @xmath129 .
the boundary conditions on the inward and outward beams are @xmath130 at @xmath59 and @xmath131 at @xmath37 . as in sect .
2.7 , the line profile @xmath103 is given by the voigt function assuming pure radiation damping .
but in view of the moderate optical depths of the subordinate transitions , a microturbulent contribution of 3 km s@xmath29 is now included in the doppler parameter .
a simple iterative procedure is adopted to obtain mutually consistent solutions of the equations of statistical equilibrium and line transfer .
eq ( 5 ) is first solved with @xmath132 , its optically - thin limit .
the resulting level populations are then used to obtain improved values of @xmath133 from the transfer equation .
this process is repeated until the average of the absolute fractional changes of the level populations in the atmosphere are @xmath134 .
when the iterations have converged , each line s emergent flux is @xmath135 , from which its residual intensity @xmath136 and equivalent width @xmath137 are readily computed . with only minor modifications ,
the code described above can also be used to compute the h line spectrum of the irradiated atmosphere of com j1740 - 5340 .
the vector @xmath90 already contains elements giving the fractions of @xmath138 used to excite and ionize h atoms . accordingly , with the standard assumption of in situ re - absorption of lyman- continuum and line photons , the h line spectrum is similarly obtained by iteratively solving the statistical equilibrium and transfer equations of sects .
2.6 and 2.8 .
one point of difference is that impact ionizations of h are allowed to increase @xmath51 from its lte value .
in this section , the simple treatment of line formation in an irradiated atmosphere developed in sect.2 is applied to compute the strengths of he lines in the spectrum of com j1740 - 5340 . with atmospheric parameters fixed ( sect .
2.3 ) , line strengths depend on @xmath139 and @xmath21 . selecting a typical point on the disk , we set @xmath140 and thus limit the investigation to exploring the atmosphere s response to a beam with conversion efficiency @xmath26 and effective photon energy @xmath21 .
line strengths depend on @xmath21 only through the function @xmath141 defined in sect .
2.4 . from the monte carlo calculations made in determining this function , it is informative to compute the fraction of incident flux deposited in the reversing layer , the fraction deposited below the photosphere at @xmath37 , and the fraction emerging back into space at @xmath59 . fig . 1 plots these deposited fractions as a function of @xmath21 for @xmath140 .
the behaviour of these curves is readily understood from the physics of compton scattering . at low energies ,
the cross section is approximately that of thomson scattering and a negligible fraction of the photon s energy is transferred to the compton electron . as a result ,
most of the incident flux is simply reflected back into space , with only a small fraction deposited within the reversing layer . at high energies , on the other hand , the greatly reduced klein-
nishina cross section allows a large fraction of the incident photons to pass right through the reversing layer , so that most of the incident flux is thermalized below the photosphere .
the local @xmath142 is therefore increased and , in consequence , so also are temperatures in the reversing layer . in this way , a sufficiently strong beam of @xmath2-rays with @xmath143 has an _ indirect _ effect on the observed line spectrum , with the @xmath144 levels of he@xmath16 being populated by the photospheric continuum and by collisions with thermal electrons .
this is the circumstance envisaged by ferraro et al .
( 2003 ) . however , for a broad range of intermediate photon energies @xmath145 , we see from the bell - shaped curve in fig .
1 that a significant fraction of the incident flux is deposited in the reversing layer ( @xmath146 ) and thus _ directly _ affects line formation through impact excitations and ionizations . fig .
1 shows that non - thermal excitation of the he@xmath0i lines is favoured when the beam s photon energy is @xmath147 1 mev .
accordingly , we now set @xmath148 1 mev , since a definitive failure at this @xmath21 would be fatal for this interpretation . with @xmath21 fixed , the single remaining parameter is @xmath26 , the conversion efficiency .
we now take @xmath26 to be a free parameter , postponing till later a discussion of observational and theoretical constraints on its magnitude ( sect .
as @xmath26 increases from zero , the impact rates @xmath149 in eq.(5 ) increase proportionately - see eq.(4 ) , resulting eventually in column densities for the @xmath144 levels sufficient for detectable absorption of the photospheric continuum . the simplest way of illustrating this effect is by constructing curves of growth with @xmath26 as the abscissa and dimensionless equivalent width @xmath150 as ordinate .
accordingly , for @xmath151 ranging from -5 to -0.5 , the coupled radiative transfer and statistcal equilibrium problems of sect .
2 have been solved and the resulting line profiles and equivalent widths @xmath152 computed for all transitions represented in the adopted atomic model for he@xmath16 . fig .
2 is the curve of growth for the line he@xmath0i @xmath15876 .
we see that @xmath152 increases linearly with @xmath26 for @xmath153 and thereafter more slowly .
this is reminiscent of conventional curves of growth , where the linear part is followed by a flat part as the lines doppler cores become saturated while their damping wings remain optically thin .
but here there is the additional effect of the partial filling - in of the absorption line by intrinsic line emission within the reversing layer .
these additional line photons are created in the radiative cascades following impact ionizations and excitations .
the role of intrinsic emission in reducing @xmath152 can be quantified approximately by also computing , for the given level populations , the line profile under the assumption of coherent scattering . for the adopted schuster - schwarzschild reversing layer ,
the residual intensity is then @xmath154 , where @xmath155 is the column density of absorbing atoms and @xmath156 is given by eq.(6 ) .
the curve of growth thus obtained is the dashed curve in fig.2 .
the vertical difference between the two curves increases with @xmath26 , reflecting the growing contribution from cascades .
the actual line profiles used to derive these curves of growth are plotted in fig . 3 for @xmath157 .
in this case , intrinsic emission reduces @xmath152 from 0.35 to 0.28 .
since the strengths of other optical he@xmath0i lines might also be relevant to testing this excitation mechanism , a selection of their curves of growth are plotted in fig.4 . as with the @xmath15876line ,
all lines have curves of growth that eventually depart from linearity in reponse to the combined effects of core saturation and intrinsic emission .
the most extreme case is that of the @xmath16678 line , whose @xmath152 increases to a maximum of @xmath158 at @xmath159 and then declines precipitously , going into emission for @xmath160 .
in this section , the predictions of the non - thermal excitation model are compared with observational data for com j1740 - 5340 . the prediction that each line has a maximum @xmath152 offers an immediate prospect of observational contradiction . from fig.2
, we see that @xmath161 reaches a maximum of @xmath162 at @xmath163 . for comparison ,
observed values can be estimated from the spectra published by sabbi et al .
( 2003 ; fig.4 ) .
these spectra show that @xmath152 varies with phase but reaches strengths of @xmath164at phases 0.02 and 0.56 .
this comparison shows that the predicted maximum exceeds observed values with a comfortable safety margin .
this is just as well since , at these phases , a significant fraction of the companion s projected disk is not irradiated by the beam and is thus not expected to contribute to the integrated line profile .
eventually , when higher s / n spectra are obtained , it will be worthwhile to calculate line profiles integrated over the disk .
this will not only give more definitive tests of line strengths but also allow the phase - dependent velocity displacement of the line centre from the orbital velocity of com j1740 - 5340 to be tested .
such displacements are a consequence of only one hemisphere of the synchronously rotating companion ( ferraro et al .
2003 ) being irradiated .
of course , with data of exceptional quality , the technique of doppler tomography could be used to make inferences about the pulsar s beam pattern . from fig.2 , we find that @xmath161 is comparable with the observed strengths for @xmath165 . but
this assumes the beam sweeps out solid angle @xmath34 steradian ( sect .
this must be increased by a factor @xmath166 for the entire facing hemisphere to be irradiated .
thus , more realistically , the required efficiency is @xmath167 .
the next question , therefore , is whether this is achievable for photon energies @xmath168 mev .
no data relevant to this question has yet been acquired for psr j1740 - 5340 .
but such data is available for the binary millisecond pulsar psr j0218 + 4232 , whose spin period ( 2.3ms ) and spin - down luminosity ( @xmath169 erg s@xmath29 ) ( mineo et al .
2000 ) are not too different from those of psr j1740 - 5340 .
detections and upper limits for photon energies from 0.05 - 10,000 mev are available for this pulsar . from this data ,
kuiper et al . ( 2000 ) suggest that this pulsar s maximum luminosity density is reached in the comptel mev range just below the reported upper limits . accepting this suggestion and thus interpreting the 2@xmath170 upper limit reported for the comptel 0.75 - 3 mev band as a detection ,
we find @xmath171 15% for a 1 steradian beam .
this is to be compared with @xmath171 7% found by kuiper et al . from egret
_ detections _ for energies between 100 mev and 1 gev .
this interpretation of the data for psr j0218 + 4232 indicates that the efficiency required if @xmath1721 mev presents no problem .
but note that , for this same data set , dyks & rudak ( 2002 ) offer a theoretical interpretation which predicts fluxes in the comptel bands far below the reported upper limits . specifically , at @xmath31 mev their model predicts @xmath173 , corresponding to a barely detectable line at @xmath15876 .
the 1-steradian efficiency @xmath174 needed to produce the observed strength of the @xmath15876 line corresponds to an incident @xmath2-ray flux @xmath175erg @xmath36 s@xmath29 .
even if all of this were thermalized below the photosphere , the increase in @xmath142 of the irradiated hemisphere is only @xmath176k .
thus the proposed explanation of the @xmath15876 line is consistent with the absence of a detectable photometric signature of heating ( sect .
hydrogen line profiles have also been calculated as a function of @xmath26 .
the initial thought was that the observed strength of the @xmath15876 line might be achieved simultaneously with h@xmath177 going into emission , thus explaining the star s earlier listing as a by dra candidate ( taylor et al .
2001 ) . in the event , at values of @xmath26 corresponding to @xmath15876reaching its observed strength ,
the contribution of intrinsic emission to h@xmath177 is still neglible .
however , this is not a failure of the model since it is clear from the spectra published by sabbi et al .
( 2003 ) that h@xmath177 emission originating close to the surface of com j1740 - 5340 is _ not _ due to irradiation .
this conclusion follows because the h@xmath177 emission remains stong even when most of the facing hemisphere is hidden from the observer .
most likely , as sabbi et al .
conclude , the emission has a chromospheric origin as commonly found in short - period , low mass binaries .
the above discussion has focussed on he@xmath0i @xmath15876 since this the line from which ferraro et al .
( 2003 ) and sabbi et al .
( 2003 ) infer that the pulsar s beam is highly anisotropic . of the other he@xmath0i lines , the one
whose predicted strength is most likely to be in conflict with observation is the @xmath16678 line , which is the analogue for the singlets of the triplet line @xmath15876 .
ferraro et al . ( 2003 ) report the presence of the @xmath16678 line but not its strength .
according to these calculations , its maximum equivalent width is only @xmath178 , which is probably inconsistent with detetection .
subsequent to the original submission of this paper , e.sabbi ( bologna ) has confirmed that the @xmath16678 line is indeed reliably detected .
moreover , it reaches an equivalent width well in excess of the above maximum .
specifically , at phase 0.56 , she reports that @xmath179 , almost an order of magnitude larger than the predicted maximum
. the weakness of @xmath16678 in these calculations contrasts with its strength in models for type ib sne ( paper i , fig.1 ) .
the origin of this difference is chemical composition .
the atmospheres of type ib sne are hydrogen - free , so there is no loss of population from singlet levels due to the ionization of hydrogen by photons emitted by the permitted ground state transitions @xmath4 ( sect .
this explanation for the predicted weakness of the @xmath16678 line is confirmed by recomputing the model at @xmath180 but now with the line escape probabilities of sect .
2.7 @xmath181 .
the result is a dramatic increase in @xmath182 from 0.022 to 0.33 .
interestingly , in this experiment @xmath161 also increases , from 0.26 to 0.35 .
the reason for this is the removal of the population decrement of the @xmath144 singlet levels relative to the corresponding triplet levels . in consequence
, there is no longer a significant net collisional transfer of excitation from the triplets to the singlets . in his report
, the referee f.r.ferraro ( bologna ) generously included unpublished information on other he lines .
this data , as also that from e. sabbi discussed above , is from the spectra described by ferraro et al .
( 2003 ) and sabbi et al .
( 2003 ) .
the triplet line @xmath13889 has a predicted strength close to that of the @xmath15876 line - see fig.4 - and so should be seen were it not blended with the h8 balmer line .
the triplet line @xmath14471 has a predicted equivalent width of 0.06 when @xmath161 = 0.19 , as observed at phases 0.02 and 0.56 .
though weak , a line of this strength should still be visible , but ferraro and colleagues finds no evidence of it .
this is undoubtedly a difficulty but one unlikely to be specific to the non - thermal mechanism .
the reason for this is that the @xmath15876 and @xmath14471 lines have the same lower level and the ratio of their predicted strengths is approximately equal to the ratio of their f - values .
thus a similar ratio should be expected for the thermal mechanism proposed by ferraro et al .
the aim of this paper has been to investigate the possibility that the formation mechanism for he@xmath0i absorption lines in the spectrum of com j1740 - 5340 is similar to that operating in the atmospheres of type ib sne . this alternative merits serious consideration in view of the powerful and remarkable constraint on the pulsar s radiation pattern that follows if the he@xmath0i lines are indeed formed by conventional thermal processes .
the result of this investigation is moderately discouraging for the non - thermal interpretation but not yet a decisive rejection . on the positive side , the crucial line @xmath15876 achieves the observed strength at seemingly feasible values of the conversion efficiency @xmath26 .
moreover , the corresponding heating effect is consistent with the non - detection of any such effect photometrically .
however , despite these successes , the strengths of other he@xmath0i lines , especially @xmath16678 , present difficulties .
the question then is whether such difficulties are fatal or can be overcome by refinement or modification of the simple model used here . among refinements to be considered
are the effect of the beam on the continuum opacity , specifically the dissociation of the h@xmath50 ion by impacts ( geltman 1960 ) , the ionization of surface layers by impacts , and the inclusion of an empirical model for the chromosphere discovered by sabbi et al .
a higher level treatment of a stellar atmosphere s response to irradiation by gamma rays is an attractive technical challenge and likely to have application to other close companions of exotic objects .
nevertheless , given the relatively low s / n of the spectra of com j1740 - 5340 and the poorly constrained parameters of the irradiation model , it is perhaps doubtful that a decision between the thermal and non - thermal line formation mechanisms can soon be reached purely on the basis of line diagnostics .
accordingly , the prediction by sabbi et al .
( 2003 ) that up to 1% of the disk of com j1740 - 5340 radiates with @xmath183k should be tested with deep uv imaging at @xmath184 .
a detection of a photometric signature of a heating effect at such wavelengths would be a decisive confirmation of the thermal mechanism and therefore of a narrowly - confined pulsar beam .
|
the he@xmath0i @xmath15876 absorption line recently discovered in the spectrum of the companion to the millisecond pulsar psr j1740 - 5340 is tentatively attributed to electron impact excitations due to the irradiation of its atmosphere by @xmath2-rays emitted by the pulsar s magnetosphere .
numerical calculations , similar to those carried out previously for type ib sne , indicate that a pulsar beam with photon energies @xmath3 1 mev gives rise to a @xmath15876 line of the observed strength if the beam s spin - down conversion efficiency approaches 1% . however , a significant difficulty for the proposed mechanism is the strength of the singlet line at @xmath16678 .
compared to the corresponding triplets , singlet lines are weak because of the loss of excitation when photons emitted in the series @xmath4 ionize hydrogen atoms , an effect absent in the hydrogen - free atmospheres of type ib sne .
|
the gray - scott model @xcite is a widely - studied model of a pair of reactions involving cubic autocatalysis .
it has been applied in reaction - diffusion models in one @xcite , two @xcite and three @xcite dimensions .
widely - known results include the existence of stable single spots , self - replication of spots , spontaneous formation of stripes and hexagonal arrays of spots , and turing patterns @xcite .
findings that show a lack of moving stable patterns , e.g. @xcite apply only in 1-d or a limited region of the 2-d system parameter space ( @xcite p. 81 and
@xcite p. 3 ) .
@xcite shows existence of stable multi - spot patterns within a finite domain .
the model equations are : @xmath1 this paper concerns the two - dimensional case with periodic boundary conditions .
i use similar terminology and symbols to those in @xcite : @xmath2 and @xmath3 are the concentrations of two reactants @xmath4 and @xmath5 , normalized to dimensionless units .
parameters @xmath6 and @xmath7 represent the feed rate and removal rate of the reactants in the original homogeneous continuously - stirred tank reactor model ; in diffusion systems they are typically the rates at which @xmath4 and @xmath5 permeate through a membrane separating a homogeneous supply from the gel in which the patterned reactions occur @xcite .
@xmath8 and @xmath9 are the diffusion rates of the two reactants through the gel or other medium ; considered together as a ratio @xmath10 , they constitute a third parameter that determines certain characteristics of the parameter space @xcite . in this paper , @xmath11 and @xmath12 . since i am measuring the size , velocity , and other related statistics of certain features in observed patterns
, there is a need for time and length units . for this purpose
i use the dimensionless units of length and time implicit in the equations , and refer to them as / ( `` length unit '' ) and / ( `` time unit '' ) respectively .
i used numerical simulation similar to that of pearson @xcite ( discrete euler forward integration , also called forward - time centered - space @xcite ) .
all figures shown here were produced with a @xmath13 grid , with @xmath14 representing a system size of @xmath15 , periodic boundary conditions , and @xmath16 .
results were verified using higher grid resolution , smaller time steps , and 2^nd^ and 4^th^ order runge - kutta integration . because many of these patterns exhibit novel properties ( such as stability combined with motion ) it was necessary to perform a thorough series of measurements to isolate the legitimate `` real '' phenomena from any effects that arise from simulation error . in these tests ,
the grid resolution @xmath17 , time step @xmath18 , and stability index @xmath19 ( the variable component of the courant condition , see @xcite ) were each decreased in steps of @xmath20 , @xmath21 and @xmath20 respectively .
the finest resolution used was @xmath22 with double - precision arithmetic . ' '' ''
.measurements of the velocities of the two moving patterns in figure [ fig : pi - valu ] , performed with progressively increasing grid and time resolutions .
all tests were performed in ieee double precision ; @xmath23 , @xmath24 , and gridsize @xmath25 or larger .
procedure for velocities : 1 .
wait for moving pattern to assume its steady - state appearance .
2 . locate a grid point on the centerline of the moving feature and on its leading edge , where @xmath26 is high and @xmath2 is close to 0.42 ; note its location and the elapsed system time ( in / and / respectively ) .
3 . wait for the pattern to travel at least @xmath27 .
4 . measure again , compute @xmath28 .
measurement error is 1 part in 200 .
procedure for last column : 1 .
@xmath23 ; @xmath7 set to trial value ; grid contains a single pattern .
2 . at periodic intervals ,
compute sum of all @xmath2 values in the grid .
3 . examine the @xmath2 sums to identify its asymptotic behavior : a stable moving approaches a finite asymptote and continues moving ; unstable patterns stop moving or grow without bound .
4 . repeat for a new @xmath7 value ; continue until the range of valid @xmath7 values is known to the desired precision .
@xmath29 = did not test .
[ tab : accuracy ] [ cols="^,^,^,^,^,^ " , ] these tests showed no qualitative difference in the results , but did enable more accurate determination of certain measurements .
for example , the `` measured velocities '' in tables 1 and 2 have a measurement error of 1 part in 200 due to the procedure used , and the increases in @xmath17 and @xmath18 resolution had no significant effect on the measured velocities : but the double - precision tests show a somewhat greater consistency ( i.e. lower variance ) in the same measurement across different grid resolutions .
all phenomena reported in the remainder of this paper appear at the grid resolutions of @xcite , @xmath30 , with single - precision arithmetic .
a few grid effects are apparent at this lower resolution , for example the two - spot pattern in figure [ fig : pi - valu ] reorients to a diagonal alignment .
such effects diminish sharply at higher grid resolutions .
the initial state was created in several ways .
most results were produced by starting with a background level of @xmath2 and @xmath3 set to a homogeneous state computed from @xmath6 and @xmath7 .
there is always a stable state @xmath31 , sometimes called the `` red state '' @xcite . for @xmath32
sufficiently small there are two other homogeneous states , one of which can be stable ( called the `` blue state '' , due to the color of the ph indicator bromothymol blue the laboratory experiments of @xcite ) . transforming the variables and units from muratov and osipov@xcite ,
the third homogeneous state @xmath33 exists when @xmath34 , with @xmath2 and @xmath3 given by @xmath35 where @xmath36 is @xmath37 .
starting with a grid filled with these values of @xmath2 and @xmath3 or with @xmath38 , a number of rectangles were added whose width , height , locations , and number were determined randomly , and then filled with random levels of @xmath2 and @xmath3 . for most tests , these rectangles ranged from @xmath39 to @xmath40 in area , and the frequency of occurrence of any given size rectangle was inversely proportional to its area .
density of these rectangles ranged anywhere from 1 to 40 per @xmath41 .
a typical example is shown in figure [ fig : uskate - generation ] .
0.2 . in the first image
, each rectangle has a randomly chosen @xmath2 and @xmath3 between 0 and 1 , and the background is @xmath42 .
coloring is the same as in figure [ fig : types - nx ] .
the exaggerated contrast makes many @xmath2 values completely black or white .
[ fig : uskate - generation ] , title="fig : " ] 0.2 . in the first image ,
each rectangle has a randomly chosen @xmath2 and @xmath3 between 0 and 1 , and the background is @xmath42 .
coloring is the same as in figure [ fig : types - nx ] .
the exaggerated contrast makes many @xmath2 values completely black or white .
[ fig : uskate - generation ] , title="fig : " ] 0.2 . in the first image
, each rectangle has a randomly chosen @xmath2 and @xmath3 between 0 and 1 , and the background is @xmath42 .
coloring is the same as in figure [ fig : types - nx ] .
the exaggerated contrast makes many @xmath2 values completely black or white .
[ fig : uskate - generation ] , title="fig : " ] 0.2 . in the first image ,
each rectangle has a randomly chosen @xmath2 and @xmath3 between 0 and 1 , and the background is @xmath42 .
coloring is the same as in figure [ fig : types - nx ] .
the exaggerated contrast makes many @xmath2 values completely black or white .
[ fig : uskate - generation ] , title="fig : " ] a repeatable pseudo - random algorithm was used to generate all starting patterns , and the initial seed values saved so the same simulation sequence could be reproduced at will @xcite .
the simulation was run for anywhere from @xmath43 to @xmath44 / , as needed for the phenomena under test . a complete survey of the parameter space in figure [ fig : param - space ]
was performed at intervals of 0.02 in @xmath7 and 0.04 in @xmath6 ; many starting patterns were tried at each parameter value combination .
the interested reader can find a gallery of images online at @xcite .
although all of the patterns shown here arose naturally from such initial random states , a number of techniques were used to make exploration and discovery more practical : selectively removing unwanted patterns by setting portions of the grid to the homogeneous state ; combining parts of patterns to create others ; changing parameter values ; moving small parts of the grid to test pattern integrity ; then continuing the simulation after making any of these changes .
because of their stability , each type @xmath45 pattern naturally evolves from any starting pattern that superficially resembles it , provided the dimensions and levels of @xmath2 and @xmath3 are approximately correct .
i found it exceptionally useful to visualize @xmath2 and @xmath46 simultaneously via appropriate color mappings , with the latter greatly amplified as in figure [ fig : pi - deriv ] .
an interactive simulation tool was essential for the discovery of stable moving patterns `` in the wild '' .
most work was performed on an 8-core xeon workstation ; verification tests were also performed on a powerpc workstation .
the simulation was implemented in c , divided into anywhere from 2 to 16 execution threads via pthreads .
the data set was partitioned in one dimension only ( stripes , each overlapping by two rows with the two neighbor stripes ) ; all inter - thread data exchange was via shared memory .
additional insight was gained from 1-d and 3-d numerical simulations ( the latter from the website at @xcite ) .
i noted all of the pattern types reported in pearson @xcite , including several parameter values at which two or more of pearson s types coexist ; many examples are exhibited at @xcite .
in addition i observed three new pattern types , illustrated in figure [ fig : types - nx ] and named @xmath47 , @xmath48 and @xmath45 to extend pearson s classification letters .
type @xmath47 is found throughout a large part of the area labeled / in @xcite , bordering on the regions @xmath49 , @xmath50 , and @xmath51 .
here we have stable spots ( called solitons or autosolitons by others @xcite ) that do not multiply .
these spots are static only in isolation .
pairs and groups drift apart from one another , at a rate that diminishes sharply with distance .
the velocity of two spots drifting apart in a sufficiently large and otherwise empty domain is modeled fairly well by an exponentially decaying rate : velocity @xmath52 where @xmath53 is inter - spot distance , @xmath54 and @xmath55 constant . for @xmath56 as shown here , @xmath57 : the spots drift about half as fast with every added @xmath58 of inter - spot distance . @xmath55 increases with @xmath7 and with @xmath6 ; for example , at @xmath59 , @xmath60 ; and at @xmath61 , @xmath62 . given enough time and a small domain size , motion ceases with spots roughly equidistant ; such multi - spot patterns are the subject of @xcite . at lower @xmath6 values , particularly for @xmath63 , spot - like starting patterns produce spots that oscillate with a characteristic frequency varying slightly with @xmath7 and @xmath6 , and a damping rate that diminishes sharply as @xmath32 approaches the right edge of the type @xmath47 region in figure [ fig : param - space ] . near this frontier
the stable spot becomes smaller in diameter and the central peak has @xmath64 farther from @xmath65 .
also in figure [ fig : param - space ] we see that at higher @xmath6 values type @xmath47 occurs on both sides of the saddle - node bifurcation line . as mazin et .
al . found in one dimension @xcite , at higher @xmath6 values the red state overpowers the blue state in a substantial band to the left of the saddle - node line ; however it is not strong enough to extinguish small spots .
for this reason , parameters in this band support solitons that do not grow to fill the space with the blue state .
type @xmath48 is found in a modest - sized area of the parameter space which includes @xmath66 and @xmath67 .
as shown in figure [ fig : param - space ] , type @xmath48 appears on both sides of the saddle - node bifurcation .
it also appears on both sides of the subcritical / supercritical boundary at @xmath68 ( when @xmath69 , the model without diffusion oscillates indefinitely ; see @xcite and @xcite ) .
type @xmath48 patterns very closely resemble the b - z ( belousov - zhabotinsky ) reaction in a petri dish .
an initial pattern containing a traveling wave front with free ends produces two spiral seeds and usually results in sustained activity . at some parameter values
the waves develop flaws similar to the segmented waves shown in @xcite ( which is discussing a cdima system ) ; these flaws contribute to the production of more spiral seeds .
occasionally , double spirals also occur .
a self - sustaining population of spiral seeds is needed to maintain the pattern , and the density of seeds varies with @xmath7 and @xmath6 .
the longevity of the pattern is highly dependent on the size of the domain . at @xmath70 , twenty starting patterns like that in figure [ fig : uskate - generation ] ( a @xmath71 domain ) had an average life - span of 5400 / before all waves died out .
twenty trials in a @xmath72 domain with starting patterns of the same density all lasted longer than @xmath73 /.
( 11.0 , 11.0)(0,0 ) ( 1.0 , 0.0 ) coordinate(a ) ( 1.0 , 11.0 ) coordinate(b ) ( 0.0 , 1.0 ) coordinate(c ) ( 11.0 , 1.0 ) coordinate(d ) ; ( a ) ( b ) ; ( c ) ( d ) ; ( 3.0 , 1.0 ) ( 3.0 , 11.0 ) ; ( 3.0 , 0.8 ) ( 3.0 , 1.0 ) ; ( 2.6 , 0.4)@xmath74 ; ( 5.0 , 1.0 ) ( 5.0 , 11.0 ) ; ( 5.0 , 0.8 ) ( 5.0 , 1.0 ) ; ( 4.6 , 0.4)@xmath75 ; ( 7.0 , 1.0 ) ( 7.0 , 11.0 ) ; ( 7.0 , 0.8 ) ( 7.0 , 1.0 ) ; ( 6.6 , 0.4)@xmath76 ; ( 9.0 , 1.0 ) ( 9.0 , 11.0 ) ; ( 9.0 , 0.8 ) ( 9.0 , 1.0 ) ; ( 8.6 , 0.4)@xmath77 ; ( 1.0 , 3.0 ) ( 11.0 , 3.0 ) ; ( 0.8 , 3.0 ) ( 1.0 , 3.0 ) ; ( 0.0 , 2.9)@xmath78 ; ( 1.0 , 5.0 ) ( 11.0 , 5.0 ) ; ( 0.8 , 5.0 ) ( 1.0 , 5.0 ) ; ( 0.0 , 4.9)@xmath74 ; ( 1.0 , 7.0 ) ( 11.0 , 7.0 ) ; ( 0.8 , 7.0 ) ( 1.0 , 7.0 ) ; ( 0.0 , 6.9)@xmath76 ; ( 1.0 , 9.0 ) ( 11.0 , 9.0 ) ; ( 0.8 , 9.0 ) ( 1.0 , 9.0 ) ; ( 0.0 , 8.9)@xmath79 ; ( 10.3 , 0.5)@xmath7 ; ( 0.3 , 10.2)@xmath6 ; ( 0.524555 , 1.400000 ) ( 1.071068 , 1.500000 ) ( 1.545967 , 1.600000 )
( 1.966600 , 1.700000 )
( 2.344272 , 1.800000 )
( 2.686833 , 1.900000 )
( 3.000000 , 2.000000 )
( 3.288088 , 2.100000 )
( 3.554451 , 2.200000 )
( 3.801754 , 2.300000 )
( 4.032160 , 2.400000 )
( 4.247449 , 2.500000 )
( 4.449111 , 2.600000 )
( 4.638405 , 2.700000 )
( 4.816408 , 2.800000 )
( 4.984049 , 2.900000 )
( 5.142136 , 3.000000 )
( 5.291377 , 3.100000 )
( 5.432397 , 3.200000 )
( 5.565751 , 3.300000 )
( 5.691933 , 3.400000 )
( 5.811388 , 3.500000 )
( 5.924515 , 3.600000 )
( 6.031677 , 3.700000 )
( 6.133201 , 3.800000 )
( 6.229386 , 3.900000 )
( 6.320508 , 4.000000 )
( 6.406817 , 4.100000 )
( 6.488544 , 4.200000 )
( 6.565902 , 4.300000 )
( 6.639089 , 4.400000 )
( 6.708287 , 4.500000 )
( 6.773666 , 4.600000 )
( 6.835384 , 4.700000 )
( 6.893589 , 4.800000 )
( 6.948418 , 4.900000 )
( 7.000000 , 5.000000 )
( 7.048457 , 5.100000 )
( 7.093902 , 5.200000 )
( 7.136441 , 5.300000 )
( 7.176177 , 5.400000 )
( 7.213203 , 5.500000 )
( 7.247611 , 5.600000 )
( 7.279483 , 5.700000 )
( 7.308902 , 5.800000 )
( 7.335944 , 5.900000 )
( 7.360680 , 6.000000 )
( 7.383180 , 6.100000 )
( 7.403509 , 6.200000 )
( 7.421729 , 6.300000 )
( 7.437900 , 6.400000 )
( 7.452079 , 6.500000 )
( 7.464319 , 6.600000 )
( 7.474673 , 6.700000 )
( 7.483189 , 6.800000 )
( 7.489916 , 6.900000 )
( 7.494897 , 7.000000 )
( 7.498178 , 7.100000 )
( 7.499799 , 7.200000 )
( 7.499801 , 7.300000 ) ( 7.498221 , 7.400000 )
( 7.495098 , 7.500000 )
( 7.490465 , 7.600000 )
( 7.484358 , 7.700000 )
( 7.476810 , 7.800000 )
( 7.467851 , 7.900000 )
( 7.457513 , 8.000000 )
( 7.445825 , 8.100000 )
( 7.432816 , 8.200000 )
( 7.418512 , 8.300000 )
( 7.402941 , 8.400000 ) ( 7.386128 , 8.500000 )
( 7.368098 , 8.600000 ) ( 7.348874 , 8.700000 )
( 7.328480 , 8.800000 )
( 7.306939 , 8.900000 )
( 7.284271 , 9.000000 )
( 7.260499 , 9.100000 )
( 7.235642 , 9.200000 )
( 7.209721 , 9.300000 )
( 7.182753 , 9.400000 )
( 7.154759 , 9.500000 )
( 7.125757 , 9.600000 )
( 7.095762 , 9.700000 )
( 7.064794 , 9.800000 )
( 7.032868 , 9.900000 )
( 7.000000 , 10.000000 )
( 6.966206 , 10.100000 )
( 6.931502 , 10.200000 )
( 6.895901 , 10.300000 )
( 6.859419 , 10.400000 )
( 6.822070 , 10.500000 )
( 6.783867 , 10.600000 )
( 6.744823 , 10.700000 )
( 6.704952 , 10.800000 )
( 6.664265 , 10.900000 ) ; ( 0.375156 , 1.700000 )
( 0.773589 , 1.800000 )
( 1.143118 , 1.900000 )
( 1.487850 , 2.000000 )
( 1.810956 , 2.100000 )
( 2.114936 , 2.200000 ) ( 2.401812 , 2.300000 )
( 2.673245 , 2.400000 )
( 2.930624 , 2.500000 )
( 3.175121 , 2.600000 )
( 3.407739 , 2.700000 )
( 3.629344 , 2.800000 )
( 3.840694 , 2.900000 )
( 4.042442 , 3.000000 )
( 4.235181 , 3.100000 )
( 4.419423 , 3.200000 )
( 4.595635 , 3.300000 )
( 4.764234 , 3.400000 )
( 4.925595 , 3.500000 )
( 5.080054 , 3.600000 )
( 5.227922 , 3.700000 )
( 5.369479 , 3.800000 )
( 5.504983 , 3.900000 )
( 5.634667 , 4.000000 )
( 5.758747 , 4.100000 )
( 5.877426 , 4.200000 )
( 5.990882 , 4.300000 )
( 6.099287 , 4.400000 )
( 6.202797 , 4.500000 )
( 6.301561 , 4.600000 )
( 6.395709 , 4.700000 )
( 6.485368 , 4.800000 )
( 6.570653 , 4.900000 )
( 6.651676 , 5.000000 )
( 6.728535 , 5.100000 )
( 6.801324 , 5.200000 )
( 6.870130 , 5.300000 )
( 6.935035 , 5.400000 )
( 6.996113 , 5.500000 )
( 7.053433 , 5.600000 )
( 7.107066 , 5.700000 )
( 7.157064 , 5.800000 )
( 7.203489 , 5.900000 )
( 7.246387 , 6.000000 )
( 7.285811 , 6.100000 )
( 7.321797 , 6.200000 )
( 7.354391 , 6.300000 )
( 7.383623 , 6.400000 )
( 7.409529 , 6.500000 )
( 7.432134 , 6.600000 )
( 7.451463 , 6.700000 )
( 7.467537 , 6.800000 )
( 7.480377 , 6.900000 )
( 7.489995 , 7.000000 )
( 7.496399 , 7.100000 )
( 7.499601 , 7.200000 ) ; ( 0.326593 , 1.500000 )
( 0.825715 , 1.600000 )
( 1.271406 , 1.700000 )
( 1.674129 , 1.800000 )
( 2.041291 , 1.900000 )
( 2.378371 , 2.000000 )
( 2.689567 , 2.100000 )
( 2.978173 , 2.200000 )
( 3.246841 , 2.300000 )
( 3.497735 , 2.400000 )
( 3.732659 , 2.500000 )
( 3.953133 , 2.600000 )
( 4.160444 , 2.700000 )
( 4.355704 , 2.800000 )
( 4.539876 , 2.900000 )
( 4.713806 , 3.000000 )
( 4.878230 , 3.100000 )
( 5.033809 , 3.200000 )
( 5.181130 , 3.300000 )
( 5.320712 , 3.400000 )
( 5.453030 , 3.500000 )
( 5.578507 , 3.600000 )
( 5.697535 , 3.700000 )
( 5.810462 , 3.800000 )
( 5.917612 , 3.900000 )
( 6.019275 , 4.000000 )
( 6.115727 , 4.100000 )
( 6.207216 , 4.200000 )
( 6.293973 , 4.300000 )
( 6.376210 , 4.400000 )
( 6.454124 , 4.500000 )
( 6.527904 , 4.600000 )
( 6.597719 , 4.700000 )
( 6.663731 , 4.800000 )
( 6.726088 , 4.900000 )
( 6.784934 , 5.000000 )
( 6.840400 , 5.100000 )
( 6.892607 , 5.200000 )
( 6.941676 , 5.300000 )
( 6.987717 , 5.400000 )
( 7.030835 , 5.500000 )
( 7.071124 , 5.600000 )
( 7.108679 , 5.700000 )
( 7.143590 , 5.800000 )
( 7.175938 , 5.900000 )
( 7.205805 , 6.000000 )
( 7.233266 , 6.100000 )
( 7.258395 , 6.200000 )
( 7.281254 , 6.300000 )
( 7.301914 , 6.400000 )
( 7.320436 , 6.500000 )
( 7.336878 , 6.600000 )
( 7.351296 , 6.700000 )
( 7.363748 , 6.800000 )
( 7.374284 , 6.900000 )
( 7.382952 , 7.000000 )
( 7.389803 , 7.100000 )
( 7.394881 , 7.200000 )
( 7.398229 , 7.300000 )
( 7.399892 , 7.400000 )
( 7.399907 , 7.500000 )
( 7.398315 , 7.600000 )
( 7.395154 , 7.700000 )
( 7.390458 , 7.800000 )
( 7.384266 , 7.900000 )
( 7.376606 , 8.000000 ) ( 7.367513 , 8.100000 ) ( 7.357018 , 8.200000 )
( 7.345152 , 8.300000 )
( 7.331942 , 8.400000 )
( 7.317418 , 8.500000 )
( 7.301605 , 8.600000 )
( 7.284531 , 8.700000 )
( 7.266221 , 8.800000 )
( 7.246699 , 8.900000 )
( 7.225990 , 9.000000 )
( 7.204115 , 9.100000 )
( 7.181097 , 9.200000 )
( 7.156959 , 9.300000 )
( 7.131719 , 9.400000 )
( 7.105401 , 9.500000 )
( 7.078023 , 9.600000 )
( 7.049604 , 9.700000 )
( 7.020160 , 9.800000 )
( 6.989714 , 9.900000 )
( 6.958280 , 10.000000 )
( 6.925876 , 10.100000 )
( 6.892519 , 10.200000 )
( 6.858224 , 10.300000 )
( 6.823010 , 10.400000 )
( 6.786888 , 10.500000 )
( 6.749874 , 10.600000 )
( 6.711983 , 10.700000 )
( 6.673228 , 10.800000 )
( 6.633624 , 10.900000 ) ; ( 6.3 , 10.7)@xmath47 ; ( 6.7 , 10.7)@xmath47 ; ( 6.7 , 10.3)@xmath47 ; ( 7.1 , 10.3)@xmath47 ; ( 7.1 , 9.9)@xmath47 ; ( 7.5 , 9.9)@xmath47 ; ( 7.1 , 9.5)@xmath47 ; ( 7.5 , 9.5)@xmath47 ; ( 7.5 , 9.1)@xmath47 ; ( 7.9 , 9.1)@xmath47 ; ( 7.5 , 8.7)@xmath47 ; ( 7.9 , 8.7)@xmath47 ; ( 8.3 , 8.7)@xmath47 ; ( 7.9 , 8.3)@xmath47 ; ( 8.3 , 8.3)@xmath47 ; ( 7.9 , 7.9)@xmath47 ; ( 8.3 , 7.9)@xmath47 ; ( 8.7 , 7.9)@xmath47 ; ( 8.3 , 7.5)@xmath47 ; ( 8.7 , 7.5)@xmath47 ; ( 8.3 , 7.1)@xmath47 ; ( 8.7 , 7.1)@xmath47 ; ( 9.1 , 7.1)@xmath47 ; ( 8.3 , 6.7)@xmath47 ; ( 8.7 , 6.7)@xmath47 ; ( 9.1 , 6.7)@xmath47 ; ( 8.3 , 6.3)@xmath47 ; ( 8.7 , 6.3)@xmath47 ; ( 9.1 , 6.3)@xmath47 ; ( 9.5 , 6.3)@xmath47 ; ( 8.3 , 5.9)@xmath47 ; ( 8.7 , 5.9)@xmath47 ; ( 9.1 , 5.9)@xmath47 ; ( 9.5 , 5.9)@xmath47 ; ( 8.3 , 5.5)@xmath47 ; ( 8.7 , 5.5)@xmath47 ; ( 9.1 , 5.5)@xmath47 ; ( 9.5 , 5.5)@xmath47 ; ( 8.3 , 5.1)@xmath47 ; ( 8.7 , 5.1)@xmath47 ; ( 9.1 , 5.1)@xmath47 ; ( 9.5 , 5.1)@xmath47 ; ( 8.3 , 4.7)@xmath47 ; ( 8.7 , 4.7)@xmath47 ; ( 9.1 , 4.7)@xmath47 ; ( 9.5 , 4.7)@xmath47 ; ( 8.3 , 4.3)@xmath47 ; ( 8.7 , 4.3)@xmath47 ; ( 9.1 , 4.3)@xmath47 ; ( 7.9 , 3.9)@xmath47 ; ( 8.3 , 3.9)@xmath47 ; ( 8.7 , 3.9)@xmath47 ; ( 7.9 , 3.5)@xmath47 ; ( 8.3 , 3.5)@xmath47 ; ( 6.548 , 9.66)@xmath45 ; ( 6.966 , 8.22)@xmath45 ; ( 6.9 , 6.9)@xmath80 $ ] ; ( 6.952 , 5.82)@xmath45 ; ( 4.3 , 2.3)@xmath48 ; ( 3.5 , 2.2)@xmath48 ; ( 1.9 , 1.9)@xmath48 ; ( 2.7 , 1.9)@xmath48 ; ( 1.5 , 1.7)@xmath48 ; c c 0.2 ( inert autosolitons ) , @xmath48 ( b - z targets and spirals ) , and @xmath45 ( stable , stationary and moving localized patterns ) .
this color key applies to all subsequent figures .
[ fig : types - nx ] , title="fig : " ] & 0.2 ( inert autosolitons ) , @xmath48 ( b - z targets and spirals ) , and @xmath45 ( stable , stationary and moving localized patterns ) .
this color key applies to all subsequent figures .
[ fig : types - nx ] , title="fig : " ] + 0.2 ( inert autosolitons ) , @xmath48 ( b - z targets and spirals ) , and @xmath45 ( stable , stationary and moving localized patterns ) .
this color key applies to all subsequent figures .
[ fig : types - nx ] , title="fig : " ] & 0.2 ( 0,0 ) rectangle ( 3.6,2.5 ) ; ( 0.1,1.7 ) rectangle ( 0.8,2.4 ) ; ( 1.0,1.9)@xmath81 ; ( 0.1,0.9 ) rectangle ( 0.8,1.6 ) ; ( 0.1,1.1 ) coordinate(a ) ( 0.1,0.9 ) coordinate(b ) ( 0.3,0.9 ) coordinate(c ) ( 0.8,1.4 ) coordinate(d ) ( 0.8,1.6 ) coordinate(e ) ( 0.6,1.6 ) coordinate(f ) ; ( a ) ( b ) ( c ) ( d ) ( e ) ( f ) cycle ; ( 0.8,0.9 ) coordinate(h ) ; ( c ) ( h ) ( d ) cycle ; ( 1.0,1.1)@xmath82 ; ( 0.1,0.1 ) rectangle ( 0.8,0.8 ) ; ( 1.0,0.3)@xmath83 ; type @xmath45 is the most novel and the subject of the rest of this paper .
i found patterns in this category for @xmath6 values ranging from @xmath74 to @xmath84 , in a very thin band of @xmath7 values ; figure [ fig : param - space ] shows four representative locations in this band .
the band runs roughly parallel to the saddle - node and turing lines , and to the left ( lower @xmath7 values ) of both .
it is also at lower @xmath7 values than the area investigated by mcgough and riley @xcite
. in @xcite pearson explored some parameter values near this area , initializing the system with red state @xmath85 and a small central rectangle of @xmath86 , and found that the system evolved to a homogeneous blue state ; these were designated /.
@xcite using a similar starting pattern in one - dimensional simulations , explored a greater number of @xmath6 and @xmath7 combinations , and found many for which stable localized structures result ; these were designated /.
at all @xmath6 values the @xmath45 band is very narrow in @xmath7 as compared to its distance from the bifurcation lines .
for example , when @xmath87 , the saddle - node bifurcation is at @xmath88 , the hopf bifurcation is at @xmath89 , the turing bifurcation is at @xmath90 , and the range of valid @xmath7 values for the object in the lower - left of figure [ fig : pi - valu ] is @xmath91 .
thus the width of the @xmath45 band near @xmath87 is about @xmath92 in units of @xmath7 , while the distance to the nearby turing region is ten times as great .
pearson s pattern type @xmath93 ( see figure 2 in @xcite ) shares some qualities with type @xmath45 .
it is in region / of mazin et .
@xcite , but its lack of solitary spots indicates it is probably slightly outside the @xmath45 band .
figure [ fig : pi - valu ] shows five patterns at @xmath23 and @xmath94 . the three in the upper row are stable non - moving patterns .
contrast has been exaggerated in the area of @xmath82 to show the concentric rings or `` halos '' that surround all patterns in systems with these parameters .
the halos are concentric stationary waves of alternating sign superimposed on the homogeneous state values @xmath95 in .
the full range of @xmath2 and @xmath3 levels for type @xmath45 patterns at @xmath96 is @xmath97 and @xmath98 .
the other two patterns in figure [ fig : pi - valu ] move to the left , indefinitely at constant speed ; the three - spot pattern moves at about 1 / per @xmath99 , and the -shaped pattern at about 1 / per @xmath100 .
all of the patterns in [ fig : pi - valu ] arise frequently from random starting patterns , and are resilient to noise and other perturbations .
if any of them is perturbed by shifting half of the pattern in any direction a distance on the order of 0.02 / , further simulation results in a return to the canonical forms shown here .
figure [ fig : pi - deriv ] shows the time derivative of @xmath2 for the same five patterns .
the motion of the two patterns in the bottom half of the figure is clear .
figure [ fig : rotate ] shows two rotating patterns .
both rotate clockwise ; the four - spot pattern performs one full revolution in about @xmath101 ; the other takes about @xmath102 .
disruptions to these will cause a momentary change in rotational velocity followed by a return to the normal rotation rate after the spots return to the stable alignment .
many stripelike patterns like those in figure [ fig : stripes ] arise from random starting patterns ; one - ended forms are more common than the two - ended versions shown here .
they are stable in the central linear section , in the direction perpendicular to their length , but are unstable in the other dimension .
any deviation from a straight line will increase , first slowly and then with increasing speed , forming meanders like those of a river in a floodplain .
these linear forms grow at both ends .
the top one is the fastest , each end grows at about 1 / per @xmath103 , slightly slower than the speed of the -shaped pattern in figure [ fig : pi - valu ] .
the second example grows at 1 / per @xmath104 from each end .
the third example has two spots at each end that make it grow ( at about 1 / per @xmath105 ) ; without these spots the pattern shrinks at each end .
when encountering other objects , these stripelike patterns will change direction , stop growing , or ( frequently in the case of the topmost example ) break down into some other form , such as separate parallel stripes .
systems that contain active growing stripes will usually grow to fill all available space with spots and stripes , but the time taken and the final proportion of spots to stripes is highly variable , and depends on tiny details of the initial configuration .
this is substantially different from the behavior of stripes at other parameter values in the gray - scott system , whose final density can generally be predicted from the parameter values alone .
the central portions of the stripelike patterns in figure [ fig : stripes ] suggest a connection with certain one - dimensional localized structures found by mazin et .
al in @xcite at parameter values in their region /.
in my tests i found that the dimensions and levels of @xmath2 and @xmath3 of features in 1-d simulations were effectively the same as cross - sections of stripe - like patterns in 2-dimensional simulations when the same parameter values are used .
0.2 to its time derivative for pattern type @xmath45 .
coloring for all but ( b ) is the same as above . in image ( b )
, white represents @xmath106 , shades of gray where @xmath107 , and black otherwise .
( c ) : two stable rotating patterns .
( d ) : three linear growing patterns .
@xmath108 for all figures ; details in section [ sec - stability ] .
[ fig : partials ] , title="fig : " ] 0.2 to its time derivative for pattern type @xmath45 .
coloring for all but ( b ) is the same as above . in image ( b ) , white represents @xmath106 , shades of gray where @xmath107 , and black otherwise .
( c ) : two stable rotating patterns .
( d ) : three linear growing patterns .
@xmath108 for all figures ; details in section [ sec - stability ] .
[ fig : partials ] , title="fig : " ] 0.2 to its time derivative for pattern type @xmath45 .
coloring for all but ( b ) is the same as above . in image ( b )
, white represents @xmath106 , shades of gray where @xmath107 , and black otherwise .
( c ) : two stable rotating patterns .
( d ) : three linear growing patterns .
@xmath108 for all figures ; details in section [ sec - stability ] .
[ fig : partials ] , title="fig : " ] 0.2 to its time derivative for pattern type @xmath45 .
coloring for all but ( b ) is the same as above . in image
( b ) , white represents @xmath106 , shades of gray where @xmath107 , and black otherwise .
( c ) : two stable rotating patterns .
( d ) : three linear growing patterns .
@xmath108 for all figures ; details in section [ sec - stability ] .
[ fig : partials ] , title="fig : " ]
the patterns in figure [ fig : pi - valu ] are shown as they appear when @xmath87 and @xmath109 . if @xmath7 is diminished below @xmath110 , the -shaped moving pattern is no longer viable and decays into a single spot .
when @xmath111 the single spot shown in the upper - left of figure [ fig : pi - valu ] is no longer stable and quickly evolves to the homogeneous state , however the other three patterns made up of spots continue to exist .
when @xmath7 is lowered below @xmath112 , the patterns in the top center and lower right vanish , but the triangular three - spot pattern remains .
below @xmath113 the triangular pattern vanishes , but a 7-spot arrangement similar to that in figure [ fig : hex - k6135 ] is still viable . the 7-spot pattern dies out when @xmath114 .
0.2 is raised progressively to three nearby values . for comparison ,
( e ) shows a turing pattern arising from low - level random noise .
@xmath23 for all figures ; details in section [ sec - vary - k ] .
[ fig : hexes ] , title="fig : " ] 0.2 is raised progressively to three nearby values . for comparison ,
( e ) shows a turing pattern arising from low - level random noise .
@xmath23 for all figures ; details in section [ sec - vary - k ] .
[ fig : hexes ] , title="fig : " ] 0.2 is raised progressively to three nearby values . for comparison ,
( e ) shows a turing pattern arising from low - level random noise .
@xmath23 for all figures ; details in section [ sec - vary - k ] .
[ fig : hexes ] , title="fig : " ] 0.2 is raised progressively to three nearby values .
for comparison , ( e ) shows a turing pattern arising from low - level random noise .
@xmath23 for all figures ; details in section [ sec - vary - k ] .
[ fig : hexes ] , title="fig : " ] 0.2 is raised progressively to three nearby values . for comparison ,
( e ) shows a turing pattern arising from low - level random noise .
@xmath23 for all figures ; details in section [ sec - vary - k ] .
[ fig : hexes ] , title="fig : " ] at @xmath115 the 5-spot pattern in figure [ fig : hex - k6090 ] is stable and moves to the left slowly . using this as the starting pattern , @xmath7
is increased to 0.06110 , and a spot forms in the center to make the 6-spot pattern in figure [ fig : hex - k6110 ] .
this pattern is also left - moving and stable .
increasing @xmath7 further to @xmath116 , a seventh spot appears ( figure [ fig : hex - k6135 ] ) , and motion stops due to the attainment of symmetry . throughout this process
the spots shift away from one another slightly to attain a new steady state , with a slightly greater inter - spot spacing with each increase in @xmath7 .
increasing @xmath7 to @xmath117 , new spots appear on all sides , quickly growing to produce the uniform hexagonal grid in figure [ fig : hex - k6150 ] .
groups of spots are no longer stable at these relatively high @xmath7 values if @xmath7 is increased more gradually from 0.06135 to 0.06150 ( for example in 15 equal steps at intervals of @xmath118 ) , some or all of the spots will have time to swell into elongated `` stripes '' of high @xmath2 . even at @xmath119 the seven - spot pattern shown in [ fig : hex - k6135 ] is barely stable , and groupings of more than 7 spots produce stripelike patterns . considered together , the range of different behaviors described thus far represents a very wide spectrum of phenomena in the relatively narrow range @xmath120 .
a similar spectrum is found at higher and lower @xmath6 values , although varying in details .
for example , the behavior shown in figure [ fig : hex - k6150 ] has been observed at several values of @xmath6 from 0.046 to 0.0652 ; at higher @xmath6 values a greater change in @xmath7 is needed to precipitate the growth of the hexagonal pattern , and it appears that at significantly higher @xmath6 the phenomenon can not be produced at all .
the -shaped moving pattern has been found to be stable at @xmath6 values from 0.0492 to 0.0876 , with @xmath7 varying as illustrated in figure [ fig : param - space ] . at @xmath121 ,
the turing effect is present . here ,
no initial spots are needed to generate a pattern of spots and/or stripes . as defined by @xcite
( see also the introductory section of @xcite ) turing patterns arise spontaneously from random noise of arbitrarily low initial amplitude , diffusion plays an active role in the destabilization of the initial state , and a characteristic wavelength ( spot size and/or stripe width ) exists that is independent of the system size .
figure [ fig : turing - k6200 ] shows a typical turing pattern in the 2-d gray - scott system ; it was produced from a starting pattern of `` white '' noise with amplitude @xmath122 superimposed on the blue state @xmath95 from .
we see similarities , such as hexagonal arrangement of spots and a similar spot size and spacing , to figure [ fig : hex - k6150 ] . however , many grain boundaries , short stripes and other differences in detail are present .
this is typical of patterns throughout the turing domain , which at @xmath87 includes @xmath7 values in the range @xmath123 ( see @xcite for a derivation of the formulas for these values ; @xmath124 is the hopf bifurcation ) .
figures [ fig : expan-6058 ] and [ fig : expan-6115 ] show two frames of a simulation in which @xmath7 was gradually increased from 0.06058 to 0.06110 over @xmath125 .
the overall length of the horizontal row of spots increases with @xmath7 , and the smaller 3-spot pattern moves slightly away from it .
the entire set of 15 spots also rotates very slowly as a unit .
similar interactions between patterns that are not in direct contact are very common .
decreasing @xmath7 back to 0.06058 over a similar time period causes the pattern to return to the state in [ fig : expan-6058 ] . in figure
[ fig : int - before ] , the -shaped feature is moving to the right and slightly downward .
it interacts weakly with the stripelike feature , then more strongly with the triangular array of spots to the right , and emerges traveling on a diagonal path towards the upper - right .
an animation of this is available at @xcite .
0.2 except as otherwise noted ; details in section [ sec - complex ] .
[ fig : spacing ] , title="fig : " ] 0.2 except as otherwise noted ; details in section [ sec - complex ] .
[ fig : spacing ] , title="fig : " ] 0.2 except as otherwise noted ; details in section [ sec - complex ] .
[ fig : spacing ] , title="fig : " ] 0.2 except as otherwise noted ; details in section [ sec - complex ] .
[ fig : spacing ] , title="fig : " ] in figure [ fig : exotic ] are six rotating patterns with a `` target '' pattern in the center .
the first three are asymmetrical ; each of these transforms into the corresponding symmetrical version ; the center shifts a bit in the process .
all rotate clockwise .
the solitary spots in the second example are pushed by the triads . in figure
[ fig : targ-2x3-spots1 ] one of the spots is stationary ; after becoming symmetrical as shown in [ fig : targ-2x(3 + 1 ) ] both spots are moving .
each of these solitary spots remains at the distance shown , about twice the radius of the brightest surrounding halo , from the annulus of the `` target '' and from the two closer spots of the triad immediately following it .
figure [ fig : four - spaceships ] shows several more moving patterns .
all move to the right .
the pattern in the upper - left is an example of many variations of the `` target '' with attached spots that arise from random starting patterns .
most remain stable and move ; symmetrical forms move on a straight path and asymmetrical forms move on a curved path .
the `` target '' by itself is also stable and does not move .
the other three all involve spots that are being `` pushed '' ahead of a larger pattern ; in all cases the larger pattern moves faster if the spots in front are removed . the asymmetrical pattern in the lower - right moves on a curved path , bending to its right as it moves forward . in the process the spot is pushed out of the way , and is left behind after about the first quarter - turn of the moving pattern s path .
an animation of this is available at @xcite .
c c c 0.11 .
@xmath108 for all figures .
[ fig : exotic ] , title="fig : " ] & 0.11 .
@xmath108 for all figures .
[ fig : exotic ] , title="fig : " ] & 0.11 .
@xmath108 for all figures .
[ fig : exotic ] , title="fig : " ] + 0.11 .
@xmath108 for all figures .
[ fig : exotic ] , title="fig : " ] & 0.11 .
@xmath108 for all figures .
[ fig : exotic ] , title="fig : " ] & 0.11 .
@xmath108 for all figures .
[ fig : exotic ] , title="fig : " ] c 0.22 .
@xmath108 for all figures .
[ fig : exotic ] , title="fig : " ]
several other authors describe spots with appearance and interaction similar to the spots shown in this paper .
schenk , et .
@xcite , studying a two - component cubic autocatalytic reaction - diffusion system , found parameter values for which there is a stable spot solution with concentric rings of alternating sign and diminishing magnitude .
for certain parameter values they found that the spots interact in a similar way to that described in this paper ; they showed some stable multi - spot arrangements similar to my figure [ fig : pi - valu ] and proposed several more .
there are some differences , for example the pattern in their figure 5(e ) ( a four - spot pattern in a configuration with unequal angles ) is not stable in the gray - scott system with parameters @xmath126 , nor at several other locations i tested throughout the @xmath45 region .
i also found the pattern in their figure 5(b ) ( three spots in a straight line ) to be stable , whereas they did not .
such differences might result from differences in the spacing and relative amplitudes of the halos . in work related to @xcite , a number of researchers including purwins et .
@xcite have identified and studied `` quasi - particles '' that can arrange themselves in a very similar manner to my figures [ fig : pi - valu ] and [ fig : hex - k6150 ] .
this can be seen clearly in figures 3 ( d ) , ( g ) , and ( h ) of the web page article by stollenwerk @xcite .
note in particular the halos around the spots in their figure 3(d ) .
these spots arose in laboratory ac gas plasma discharge experiments that have other features reminiscent of patterns seen in the two - dimensional gray - scott system .
liehr , et .
al . in @xcite , modeling a dc gas plasma discharge experiment , began with a 3-component reaction - diffusion system , then transformed the equations into a two - component form with a global integration term .
in numerical simulation ( see their figures 3 and 5 ) they achieved results similar to my figures [ fig : hex - k6135 ] and [ fig : hex - k6150 ] . note the halos and arrangement of the spots in their patterns . studying a similar 3-component system ,
schenk in @xcite shows several examples that include a _ zielscheiben - struktur _ ( target structure ) closely resembling the target patterns in my figure [ fig : exotic ] .
spots with halos and similar spot - to - spot interactions are also shown in that work . in both of these 3-component systems
, a single spot can have an intrinsic velocity and display particle - like collision behavior in addition to the static attraction and repulsion effects .
the single spot in the upper - left of figure [ fig : pi - valu ] has a set of concentric halos with progressively lower amplitude and alternating sign . in multi - spot patterns
, each spot tends to be found at a location that coincides with the first positive - sign halos of neighboring spots . in the growth of hexagonal arrays
, the new spots always appear at such locations . in figure
[ fig : pi - valu ] , the -shaped moving pattern is of similar size and shape to the three - spot pattern , and both tend to return to the canonical dimensions shown here after distortion or perturbation . in more distant alignments such as that in figure [ fig : expan-6058 ] the spots are found at distances coinciding with each other s second positive - signed surrounding halos .
this is also evident in the five - spot pattern in figure [ fig : hex - k6090 ] , which remains as shown when perturbed , rather than changing to a pentagonal ring or some other arrangement . in several instances ( for example figure [ fig : stripes ] , all of figure [ fig : spacing ] , and figure [ fig : targ-2x3 ] ) we see rows of spots and other roughly linear features , halos that parallel these features , and motion and interaction that follows the locations of these roughly linear halos .
all of the foregoing suggest that areas of high @xmath2 and low @xmath3 produce a pattern of surrounding standing waves , and that these standing waves combine in a nearly linear way to produce an effect of constructive and destructive interference , causing spots and other features to drift towards a preferred alignment .
a similar explanation exists for spot interactions in the reaction - diffusion system of @xcite .
analytical research is needed to establish a basis for this theory for the gray - scott model equations , or another explanation that can account for the observed phenomena .
inasmuch as similar effects are observed in the 1-dimensional system ( see mazin et .
al @xcite and my own work @xcite ) , analytical work can probably begin with 1-dimensional systems .
there is much opportunity for further research by numerical simulation , including rigorous statistical analysis to establish the degree of stability of the patterns in response to varying types and levels of perturbation .
the majority of the present results are at a single point in the parameter space ; there is a likelihood that changing @xmath6 and @xmath127 will yield new discoveries .
the superficial similarity ( spot shape and `` halos '' and target patterns ) and behavior ( spots condensing into a bound multi - spot state similar to `` molecules '' ) in gas plasma experiments , combined with the successful modeling ( e.g. by @xcite ) of these experiments by a reaction - diffusion system in numerical simulation , strongly suggest research to identify more connections between the present work and the gas discharge systems .
concerning the moving stable structures , there are several well - studied reaction - diffusion system models , including the brusselator model with diffusion , the three - component system of @xcite , and the oregonator model of the b - z reaction @xcite ; all sharing many or most of the features of the gray - scott system ( stripes , spots , multiple homogeneous stable states , mixed - mode patterns , ability to produce turing patterns , and varying types of movement and interaction at various parameter values ) . because of these many shared traits , it seems likely that some of these other systems can also produce stable moving localized patterns and other complex phenomena like those described in the present paper .
if such phenomena exist , they likely involve parameter values that have not been thoroughly explored , or may be confined to a very narrow region of the parameter space .
finally , this author notes that the great diversity of patterns and types of interaction displayed by the gray - scott system at these specific parameter values , combined with the inherent stability of these patterns , clearly places this system in wolfram s class 4 ( complex localized structures , sometimes long - lived ) @xcite that characterizes certain discrete cellular automata that have been shown to be capable of universal computation .
constructing large systems of interacting patterns presents several challenges for long - term stability , because most interactions cause each interacting part to shift .
however , a starting pattern containing an infinite array of elements , each of which is only used a small number of times , would appear promising .
the author acknowledges jeff mcgough for early feedback ; thanks also to joseph fineman for his copy - editing experience ; and to jonathan lidbeck for his 3-d gray - scott java applet @xcite . c. p. schenk , p. schtz , m. bode , h .-
purwins , interaction of self - organized quasiparticles in a two - dimensional reaction - diffusion system : the formation of molecules , _ physical review e _ * 57 ( 6 ) * ( 1998 ) 6480 - 6486 .
purwins , yu . a. astrov , and i. brauer , self - organized quasiparticles and other patterns in planar gas - discharge systems , in m. ding , w. l. ditto , l. m. pecora , m. l. spano ( eds . ) , proceedings of the 5th experimental chaos conference , world scientific publishing company , 2000 , pp .
3 - 14 . c. p. schenk , numerische und analytische untersuchung solitrer strukturen in zwei- und dreikomponentigen reaktions - diffusions - systemen , ( inaugural dissertation ) , westflischen wilhelms university mnster , 1999 .
+ ` http://www.uni-muenster.de/physik.ap/purwins/rd/literatur/schenk_diss.pdf `
a. w. liehr , m. bode , and h .-
purwins , the generation of dissipative quasi - particles near turing s bifurcation in three - dimensional reaction - diffusion systems , in : e. krause , w. jger , ( eds . ) , high performance computing in science and engineering 2000 , springer , 2001 , pp . 425 - 439 . c. b. muratov and v. v. osipov , spike autosolitons in the gray - scott model , _ journal of physics a : mathematical and general _ * 33 * ( 2000 ) 8893 - 8916 .
y. nishiura and d. ueyama , self - replication , self - destruction , and spatio - temporal chaos in the gray - scott model , _ physical review letters _ * 15 * ( 2000 ) 281 - 289 .
t. leppnen , computational studies of pattern formation in turing systems , doctoral thesis , helsinki university of technology , november 2004 .
j. s. mcgough and k. riley , pattern formation in the gray - scott model , _ nonlinear analysis : real world applications _ * 5(1 ) * ( 2004 ) 105 - 121 .
l. stollenwerk , pattern formation in ac gas discharge systems , at website of the workgroup on nonlinear systems and pattern formation , institute of applied physics at wwu mnster . ` http://www.uni-muenster.de/physik.ap/purwins/ac/index-en.html ` r. munafo , stable moving patterns in the 1-d and 2-d gray - scott reaction - diffusion system , talk given at rutgers university mathematical physics seminar series , fall 2010 .
slides are here : ` http://www.mrob.com/sci/talks/20101209.html `
|
i show stable , localized , single and multi - spot patterns of three classes stationary , moving , and rotating that exist within a limited range of parameter values in the two - dimensional gray - scott reaction - diffusion model with @xmath0 .
these patterns exist in domains of any size , and appear to derive their stability from a constructive reinforcement effect of the standing waves that surround any feature .
there are several common elements including a spot that behaves as a quasiparticle , a u - shaped stripe , and a ring or annulus , or a portion thereof which combine to form a great variety of stable structures .
these patterns interact with each other in a variety of ways .
there are similarities to other reaction - diffusion systems and to physical experiments ; i offer several suggestions for further research .
key words : gray - scott model , reaction - diffusion , pattern formation , numerical simulation pacs : 82.40.ck , 82.40.bj , 47.54.de , 87.18.hf ' '' '' / / / * * / * * / * *
|
the explosive growth in the number of mobile devices running the android platform has attracted the attention of hackers for the wealth of sensitive information that are usually stored on mobile devices , including phone numbers , short messages , confidential emails and correspondences , and banking information and credentials .
the availability of this information in many mass - market mobile devices makes them a desirable target for hackers , who excelled at developing a large number of mobile malicious software ( malware ) , making the security of mobile devices one of the most important and challenging areas of research .
for example , according to a report by mcafee , the total number of mobile malware continued its linear climb as it broke 8 million in the second quarter of 2015 , and increased by 17% over the first quarter of the same year @xcite .
moreover , new malware families and variants were reported to appear approximately 1 million times in the same quarter . to address this trend , antivirus ( av )
vendors analyze a large number of malware samples daily in order to prevent them from spreading widely and to guide users on disinfection and risk management by classifying malware into broad families .
mobile as well as traditional malware analysis for detection and classification falls into two broad types : static and dynamic analysis . in static analysis
, strings of bytes associated with malware samples are discovered through reverse engineering and used as a signature for identifying malicious software .
although fast and efficient , static techniques are often prone to high false positive rates due to evolution in code basis and code repacking . furthermore , additional cost of those techniques is required for reverse engineering to generate reliable and meaningful signatures .
on the other hand , dynamic and behavior based analysis aims to provide methods for effectively and efficiently extracting unique patterns of each malware family based on its behavior .
malware samples of the same family often use the same code base , provide the same functionality using the same order of behavioral events @xcite , and so on . in analyzing mobile malware , unique behavior patterns can be represented by various symbols ( e.g. , permission set , api call , and system call ) and used to identify malware families . to this end , researchers previously proposed various detection and classification methods for malware analysis based on their behavior , including permission - based , api call - based and system call - based methods .
permission - based detection methods are not efficient in classifying benign applications as benign , since relevant rule sets only focus on detecting the malware .
api call - based detection methods can not generate distinct signatures until decompilation or disassembly process is completed , which is often expensive .
system call - based detection methods can more accurately detect malicious behavior than other methods , since it is impossible to modify original functionality of system calls : malware creators always attempt to disguise malicious behavior as normal behavior .
however , proposed methods in this category mainly deal with frequency of system calls well presented in malware .
the number of invoked system calls is usually small , and most of the system calls used in malware ( e.g. , read ( ) , write ( ) ) are also observed in both benign applications , affecting the accuracy of those methods . to this end , one needs to consider more features , such as arguments in the system call and network activities , to enhance malware detection and classification via behavior profiling . to overcome the drawbacks in previous methods , we propose a feature - rich anti - malware system based on behavior profiling called andro - profiler . our proposed behavior profiling system comprises mobile devices and a remote server to facilitate profiling , and adopts profiling method in the malware analysis domain .
we exploit system calls , including their arguments provided by lkm ( loadable kernel module ) and system logs ( e.g. , sms , call , and network i / o ) provided by droidbox @xcite as feature vectors for malware characterization .
we define system calls and system logs as integrated system logs from which we directly infer behavior patterns representation using the concept of behavior profiling of bayer @xcite .
we assume that : a ) malware samples have unique malicious behavior patterns , b ) malicious behavior is determined by system calls , and c ) such system call set has influence on the behavior of the program ( malware ) .
we prepare representative behavior profile for each malware family represented by integrated system logs including system calls , their arguments , and system logs of droidbox an analysis system we utilize in this work .
we construct the behavior profile of each malware sample through its integrated system logs by executing it on an emulator .
then , by comparing the behavior profiles across samples , we can detect and classify malware samples into related families .
the main contributions of this paper are as follows : 1 .
we propose a novel anti - malware system based on behavior profiling called andro - profiler .
we classify malware by exploiting the behavior profiling extracted from integrated system logs .
our method captures the behavior profiling by converting integrated system logs into human - readable contexts , which helps analysts analyze malware intuitively .
2 . andro - profiler enables av vendors to react to many species of malicious samples by classifying and matching them with those previously detected quickly and efficiently .
our system can help detect new malware including existing malware s variants and zero - day exploits .
this is further highlighted through in - depth experiments using real - world malware samples .
our proposed method is robust , and can be extended with additional features that depict the unique behavior patterns of malware .
our method can easily employ static analysis technique to capture malicious behavior , in combination of the dynamic behavior , which is shown to outperform existing techniques in the literature .
this feature of our work is highlighted by a comparison with the prior literature , experimentally .
based on where the scan and monitoring of the mobile malware takes place , malware analysis methods are classified into three types : detection methods on the mobile device , detection methods outside the mobile device , and hybrid detection methods .
we classify the literature based on the type of the malicious behavior into permission - based and footprint - based methods .
footprint - based methods include system call - based , api call - based , decompiled code - based , and xml information - based methods .
the detection methods on mobile device scan malicious behavior patterns on the mobile device and return the analysis results to the user . however , those approaches do not consider the resource constraints on the mobile device : low computing power and limited battery life , affect their usability and user experience .
the detection methods outside mobile device execute detection algorithms on an emulator or a real device running the targeted applications , and conduct static or dynamic analysis for determining the nature of those applications .
those approaches do not need to consider resource constraints , but can not respond to new malware families quickly . to overcome the drawbacks in both approaches ,
hybrid approaches have been introduced in mobile malware analysis .
client modules deployed on mobile devices collect information related to installed applications on those devices and send the information to a remote server .
the remote server then analyzes log files using their detection algorithms of choice , while not impeding usability and user experience .
table [ tab : related ] summarizes the various malware detection or classification methods in the literature . in the following , we elaborate on some of the related works in each category .
c c c c approach & method & feature & previous works + & permission & permission & .various malware detection@xmath1classification methods in previous works . [ cols="^ " , ] andro - profiler misclassified 225 benign samples as malware .
we conducted a deep analysis to understand the high false positives with andro - profiler .
interestingly , we found that some benign samples collected user s sensitive information , which we defined as a trigger for classifying malicious applications ( e.g. , imei , device i d , uuid , latitude , and longitude ) . to understand whether other anti - malware systems and scanners considered those benign applications as malware or not , we uploaded those suspected googleplay samples to virustotal and checked scanning results of various anti - virus vendors . as a result
, we found that 110 out of the suspicious benign samples ( accounting for about @xmath2 ) were diagnosed as malware .
the high rate of misclassification of benign applications is , however , understandable given various potential reasons for such infiltration of gray area applications into the market place @xcite .
we demonstrate the effectiveness of detecting zero - day malware detection .
we define an application as a zero - day malware if it has malicious behavior and it can not be detected by av vendors . in order to verify that we had appropriately detected zero - day malware , we made 91 variant samples consisting of adwo and airpush families by leveraging adam @xcite .
all samples used as the base application for the variants are among the ones which are used in the previous experiments , and detected by virustotal as malware samples . after creating the variants , we uploaded them ( as samples ) to the virustotal , and checked scanning results of various anti - virus ( av ) vendors such as f - secure , kaspersky , clamav , and avast .
we noted that none of the submitted samples is reported as a malware when we carried out our experiment . as a result of our experiment using andro - profiler
, we found that it performed well in detecting all of the variant malware samples with @xmath3 classification accuracy on average , regardless of the update method .
our proposed system only takes 55 seconds / mb for classifying each malware ; we exclude setup time for analysis such as booting time of emulator .
the majority of this time is spent in making the behavior profile ; it takes only 0.2 seconds on average to classify malware into each family .
while the performance of our system is operationally reasonable , our system is scalable both horizontally and vertically by design .
horizontally , and given that our server side components are run in a virtual environment , one can fork multiple servers by utilizing multiple virtual machines that exploit the multi - core nature of today s commodity computers .
vertically , our system can benefit from being developed in a lower level language , such as the c language , which would make the classification process run faster .
andro - profiler has a few limitations for detecting and classifying malware , since our proposed method uses integrated system logs as a feature vector and employs dynamic analysis techniques to capture malware s behavior .
first , it is difficult for our system to analyze malware that are executed only under given conditions ( e.g. , sdk version , cellular network connection status , time , or place ) .
however , this shortcoming is addressable by having various platforms tailored with various settings , as used for traditional malware in @xcite .
it is also impossible for our system to analyze malware embedding anti - malware analysis techniques .
second , our emulator - based anti - malware system is dependent on sdk version of emulator , so our approach has limitation on analyzing malicious behavior related to privilege escalation .
however , those are common drawbacks of dynamic analysis method or emulator - based detection method and addressed in the literature at some expense .
finally , our approach analyzes malware on an emulator without interaction between human and device : autonomous installation and execution .
when a malware behave upon an update or by utilizing a drive - by download attack @xcite , our approach is limited in reacting to such malware . however , autonomous installation and execution is an inevitable procedure for automation of dynamic analysis .
depending on the number of malware samples to be analyzed , one can adopt manual human interactions to analyze malware samples and vet the outcomes of the automatic classification procedure , as used in @xcite .
in this paper , we have presented andro - profiler , an anti - malware system based on behavior profiling .
using andro - profiler , we classified malware by exploiting the behavior profiling extracted from integrated system logs , which are implicitly equivalent to distinct behavior characteristics .
our behavior profiling is simple and relatively easy to understand , whereas andro - profiler is capable of distinguishing benign and malicious applications , and malicious applications into families .
furthermore , andro - profiler is capable of detecting zero - day threats , which are missed by antivirus scanners .
our experiments demonstrate that andro - profiler performs well in detecting and classifying malware families with over @xmath0 classification accuracy on average regardless of update method while crowdroid , a closely related work from the literature , performs under @xmath4 classification accuracy on average .
our experiment results indicate that it takes 55 seconds / mb to analyze a malware on average , with a lot of opportunities for improvements on scalability .
our system hence enables av vendors to react to many species of malicious samples by classifying and matching these with previous ones effectively and efficiently .
there are several directions that we will pursue in the future .
first , we would like to augment our system to not only rely on dynamic and behavioral features , but also static features that are easy to obtain from the applications at scale .
furthermore , we will explore scalability issues associated with our system by implementing some of the guidelines noted in subsection _ efficiency of malware classification_.
this research is supported by the msip ( ministry of science , ict and future planning ) , korea , under the itrc ( information technology research center ) support program ( nipa-2014-h0301 - 14 - 1004 ) supervised by the nipa(national it industry promotion agency ) .
in addition , this work is also supported by the ict r&d program of msip / iitp .
[ 14 - 912 - 06 - 002 , the development of script - based cyber attack protection technology ] . a two - page abstract on this work appeared in @xcite .
the work proposed in this paper significantly enhances the prior work , technically and content - wise , including the motivation , related - work , design , and evaluation .
riskranker : scalable and accurate zero - day android malware detection . in _ proceedings of the 10th international conference on mobile systems , applications , and services _
( 2012 ) , mobisys 12 , pp .
281294 .
andro - profiler : anti - malware system based on behavior profiling of mobile malware . in _
23rd international world wide web conference , www 14 , seoul , republic of korea , april 7 - 11 , 2014 , companion volume _ ( 2014 ) , pp .
737738 .
chatter : classifying malware families using system event ordering . in _
ieee conference on communications and network security , cns 2014 , san francisco , ca , usa , october 29 - 31 , 2014 _ ( 2014 ) , pp .
|
mass - market mobile security threats have increased recently due to the growth of mobile technologies and the popularity of mobile devices .
accordingly , techniques have been introduced for identifying , classifying , and defending against mobile threats utilizing static , dynamic , on - device , off - device , and hybrid approaches . in this paper , we contribute to the mobile security defense posture by introducing andro - profiler , a hybrid behavior based analysis and classification system for mobile malware .
andro - profiler classifies malware by exploiting the behavior profiling extracted from the integrated system logs including system calls , which are implicitly equivalent to distinct behavior characteristics .
andro - profiler executes a malicious application on an emulator in order to generate the integrated system logs , and creates human - readable behavior profiles by analyzing the integrated system logs . by comparing the behavior profile of malicious application with representative behavior profile for each malware family ,
andro - profiler detects and classifies it into malware families .
the experiment results demonstrate that andro - profiler is scalable , performs well in detecting and classifying malware with accuracy greater than @xmath0 , outperforms the existing state - of - the - art work , and is capable of identifying zero - day mobile malware samples
. * keywords . * behavior profiling , similarity , system call , android , malware
|
the detection of linear polarization at the level of @xmath4 in the optical afterglow emission of several gamma - ray bursts ( grbs ; see * ? ? ?
* for a review ) has been widely considered as a confirmation that synchrotron emission is the dominant radiation mechanism , at least in the afterglow stage .
synchrotron radiation is also believed to be the dominant emission mechanism in the prompt @xmath0-ray emission and in the emission from the reverse shock , although the observational support for this is not as strong as for the afterglow .
soon after the first detection of linear polarization in the afterglow emission @xcite it has been realized that the temporal evolution of the polarization ( both the degree of polarization @xmath5 and its position angle @xmath6 ) can probe the magnetic field structure in the emitting region , as well as the structure and the dynamics of grb jets @xcite .
while the polarization properties of the afterglow emission have received relatively wide attention @xcite , the polarization of the prompt grb emission received very little attention , in the emission from the grb ejecta , which includes the prompt @xmath0-ray emission , and the emission from the reverse shock ( the ` optical flash ' and ` radio flare ' ) . ] before the detection of a very high degree of linear polarization ( @xmath7 ) in the prompt @xmath0-ray emission of grb 021206 @xcite .
although this detection is highly controversial @xcite , it has dramatically raised the interest in the polarization properties of the prompt emission .
two main explanations have been suggested for the production of tens of percent of polarization in the prompt @xmath0-ray emission : ( i ) synchrotron emission from an ordered magnetic field in the ejecta @xcite , and ( ii ) a viewing angle just outside the sharp edge of a jet , @xmath8 @xcite , where @xmath9 and @xmath10 are the initial half - opening angle and lorentz factor of the jet , respectively
. the second explanation can work with either synchrotron emission @xcite or inverse - compton scattering of external photons @xcite .
in all the above cases , the polarization can be a good fraction of the maximal polarization for synchrotron emission , @xmath11 .
, however the averaging over the unresolved image reduces the observed polarization to values only slightly higher than those for synchrotron emission . ]
the second explanation requires a narrow jet , @xmath12 , in order to have a reasonable probability of viewing the jet at an appropriate angle , @xmath13 .
since a larger value of @xmath14 is inferred for most grbs @xcite , this would imply little or no polarization in most cases , where our viewing angle is inside the jet , @xmath15 .
the first explanation , however , would imply a high polarization in all grbs , if indeed they have a magnetic field that is ordered over angular scales @xmath16 .
the best way to reliably measure the polarization of the emission from the grb ejecta is by using the emission from the reverse shock , which includes the optical flash and the radio flare .
while no polarization measurements have been made so far in the optical flash , the polarization of the radio flare can be inferred from existing data . in [ rs ]
we discuss the existing candidates for radio flare emission in grbs and the evidence that this emission is indeed from the reverse shock . in [ obs ] we use archival data to derive upper limits on the linear and circular polarization of the radio flare emission , and in [ prop_ext ] we show that external propagation effects in the host galaxy or circumburst environment are not likely to cause a significant depolarization the radio emission .
the upper limits on the polarization are contrasted with the predictions of different theoretical models in [ theory ] . in [ prop_int ]
we discuss propagation effects inside the source .
we discuss the magnetic field configuration in the grb ejecta in light of our results in [ dis ] and give our conclusions in [ conc ] .
in order to draw conclusions regarding the magnetic field structure in the grb ejecta from the polarization measurements of the radio flares , it is important to have some confidence that the radio emission is indeed from the original ejecta , whose electrons were heated by the reverse shock and then cooled adiabatically @xcite .
the best candidates are grbs 990123 @xcite , 991216 @xcite and 020405 @xcite .
a radio flare was also reported for grb 970828 @xcite , however in this case the source was detected in the radio at only one epoch and at one frequency , and there is no good evidence that suggests that this emission arises from the reverse shock .
also , the signal to noise in this case is rather bad , and does not give meaningful constraints on the polarization .
therefore , we shall concentrate on the other three grbs for which there is better evidence that the radio flare emission is indeed from the reverse shock , and for which the quality of the data allows us to place interesting limits on the polarization . in grb 990123
the early radio emission ( up to a day or two ) has been found to agree very well with the expectations for the reverse shock emission @xcite , especially when also taking into account the prompt optical emission from this grb @xcite which is also successfully explained as emission from the reverse shock .
furthermore , it is hard to explain the early radio emission as arising from the forward shock .
therefore , in grb 990123 we have good reason to believe that the @xmath17ghz radio measurement at @xmath18days that is used in [ obs ] to derive limits on the polarization was indeed dominated by emission from the reverse shock . in grb 991216
the radio emission in the first few days is inconsistent with the forward shock emission that is responsible for the optical and x - ray emission @xcite , and a different emission component is required .
@xcite suggested either reverse shock emission , which naturally explains the early radio emission , or a two component jet model , where the early radio emission is from a spherical component with @xmath19erg running into a very low external density , @xmath20 .
such a low density is inconsistent with broad band afterglow fits to grb 991216 @xcite which give a density higher by @xmath21 orders of magnitude . also , the energy in the spherical component is very high , @xmath22erg , which is @xmath23 orders of magnitude higher than the values inferred for other grbs @xcite .
thus , the reverse shock is probably the best candidate for the early radio emission in grb 991216 , although it is hard to conclusively rule out other explanations . in grb 020405 the early rapid decay and spectral slope in the radio , @xmath24 ,
provide a reasonably good case that the early radio emission ( within the first few days ) is dominated by the reverse shock @xcite .
we use archival vla observations in order to measure the polarization of the radio flare emission discovered from grbs to date .
the vla observations are typically short observations ( @xmath2530 minutes ) that were squeezed into the vla schedule for these targets of opportunity . as such
, there is insufficient calibration information from these short runs alone to allow for good polarization calibration in the standard manner using a calibrator observed over a wide range in parallactic angle . to overcome this difficulty we combined the vla observations of the afterglows with calibration data taken from 14 days before or after the run .
the instrumental leakage terms are stable on time - scales of months , so we are confident that an optimal calibration has been obtained . as a check ,
the polarization calibration was applied to the absolute flux calibrator observed on the same day as the radio flare ( 3c286 or 3c138 ) , and the results were found to be consistent with their well - known properties .
we obtain only upper limits on the linear and circular polarization of three radio flares .
our results are summarized in table 1 .
it is worth noting that so far only upper limits have been found for the radio polarization from grb afterglows @xcite . in the best case of the bright grb 030329 ,
the @xmath1 limits are @xmath261% @xcite . &
@xmath27 & @xmath28 + & 1.49 & & @xmath29 & @xmath30 & @xmath31 + 991216 & 2.68 & @xmath32 & @xmath33 & @xmath34 & @xmath35 + & 1.49 , 2.68 & & @xmath36 & @xmath33 & @xmath37 + 020405 & 1.19 & @xmath38 & @xmath29 & @xmath39 & @xmath40
propagation effects might reduce the intrinsic linear polarization below detectable levels . a faraday screen produced by ionized
gas and magnetic fields can cause gradients in the observed polarization angle across the source , leading to depolarization if the intrinsic source size or the resolution element of the telescope is large compared to the gradients .
the rotation measures ( rm ) can be related to the line - of - sight magnetic field , @xmath41 , by @xmath42 where the upper limit of integration , @xmath43 , is the distance from the emitting source to the end of the path through the faraday screen along the line of sight . to produce a 90@xmath44 rotation at @xmath45ghz
requires a rm of @xmath46 . in our galaxy
the rm can reach up to @xmath47 at very low galactic latitudes , while at high galactic latitudes it drops to a few@xmath48 @xcite . at low galactic latitudes
@xcite find a rm difference of up to @xmath49 on scales of @xmath50 which translates to a linear scale of @xmath51pc and gradient of @xmath52 in the rm .
since in our case the source size is @xmath53pc , this would correspond to a negligible rotation measure gradient across the image , @xmath54 , some 5 orders of magnitude too low to cause significant depolarization .
the host galaxies of grbs are not thought to be greatly different from our own galaxy @xcite .
molecular clouds in our galaxy typically exhibit variation of @xmath55 in the rm over scales of @xmath56pc @xcite .
this corresponds to @xmath57 and @xmath58 across the image , which is still some 4 orders of magnitude too low to cause significant depolarization .
individual hii regions can produce enhanced rms with dispersions of order @xmath59 on scales of @xmath60pc @xcite , corresponding to @xmath61 and @xmath62 across the image , which is still some 2 - 3 orders of magnitude too low .
now we consider propagation effects in the immediate environment of the grb , which was shaped by its progenitor .
if the latter is a massive star , the immediate environment is the pre - explosion stellar wind which for a constant mass loss rate and wind velocity would give a density profile @xmath63 , at radii @xmath64 that are smaller than that of the wind termination shock .
such a density profile would imply a small deceleration radius @xmath65 and a very high density at that radius , which would in turn imply fast cooling of the reverse shock electrons and no detectable radio flare emission at @xmath66day . keeping this in mind
, we shall assume a uniform external medium . for grb 020405 , @xcite model the optical , x - ray and radio light curves with a uniform medium of density @xmath67 .
afterglow models applied to grb 020405 predict an intrinsic diameter of the radio emitting region at 1.2 days after the burst of @xmath68 cm .
values of the external density inferred from broad band afterglow modeling can be as high as @xmath69 @xcite . for grb 990123 a very low density of @xmath70 is inferred , while for grb 991216 the inferred density is more typical , @xmath71 .
magnetic fields within the circumburst medium are not expected to cause a large depolarization , unless there is a clump with a high density contrast along our line of sight .
a clump much smaller than the image size would have a small effect .
a clump larger or comparable to the image size , @xmath72 , would produce a change @xmath73 in the rotation measure across the image which is roughly independent of the clump size , @xmath43 , for a given density contrast @xmath74 .
this can be seen as follows .
we have @xmath75 and assuming flux freezing and isotropic compression , @xmath76 so that @xmath77 , and the change in the rm across the image is roughly @xmath78 , where the dependence on @xmath43 cancels out . furthermore , the gradient in the rotation measure across the image would be comparable to the total rotation measure of a clump with a size similar to that of the image , @xmath79 , causing a relative rotation of the polarization position angle of @xmath80 where @xmath81 and @xmath82 are the number density and magnetic field of the un - clumped external medium , and @xmath83 is the apparent radius of the image on the plane of the sky . thus a clump with a density contrast @xmath84 and a size @xmath85 cm is required in order to induce significant depolarization .
we also note that if such a clump lies within a distance of @xmath86pc from the progenitor star it would produce a detectable bump in the afterglow light curve , which was not observed for grbs 990123 , 991216 or 020405 .
thus , such a clump along our line of sight seems unlikely .
it is much less likely that such a clump would be along our line of sight in all the three grbs that we analyzed . instead of a single clump
, one might envision many small clumps with a covering factor of @xmath87 ( in order to cover most of the image ) .
the typical clump size would have to be @xmath88 in order to avoid producing detectable bumps in the afterglow light curve . since even for @xmath89 we need @xmath90 in order for a single clump to produce significant depolarization ( see eq . [ clump ] ) , many small clumps are required along a random line of sight .
if their rotation of the position angle would always be in the same direction , the required number of clumps would be @xmath91 .
however , since the direction of rotation of the position angle is expected to be random between different clumps , the total rotation would on average be zero with a mean r.m.s value of @xmath92 times that of a single clump .
thus a strong depolarization would require @xmath93 clumps along a random line of sight .
thus @xmath94 where @xmath95 is the total number of clumps at radii @xmath96 .
thus the total mass in the clumps would be @xmath97 while there volume filling factor would be @xmath98 .
this means that the clumps would hold @xmath99 times more mass than that of the material between the clumps .
such an extreme clumping seems highly unlikely .
the radio flare emission typically peaks on a time scale of @xmath100day after the grb .
this is usually similar to the jet break time , @xmath101 , in the afterglow light curve . at this time
the lorentz factor is typically @xmath102 , which is much smaller than its initial value , @xmath103 .
thus a good part of the jet ( or all the jet for @xmath104 ) is visible , and the observed polarization is an effective average value over this observed region .
the optical flash emission peaks at the deceleration time @xmath105 when the lorentz factor is close to its initial value ( unless the reverse shock is highly relativistic ) , and should thus have a polarization close to that of the prompt @xmath0-ray emission .
it could therefore provide information that is complimentary to that from the radio flare .
the jet break time , @xmath101 , plays an important role in the polarization light curve for most theoretical models .
therefore , it is important to know its value when comparing between theory and observations . in grb 990123
there is a jet break in the optical light curve at @xmath106days @xcite . a similar jet break time , @xmath106days ,
was found in the optical light curve of grb 991216 @xcite with a rather large uncertainty on this value . in grb 020405
the jet break is not readily apparent in the data.days .
since the most severe constraints from our polarization measurements are for an ordered toroidal magnetic field in the ejecta , in which case a larger value for the jet break time ( @xmath107days ) would imply even stricter constraints on the model , we adopt a conservative approach and use the lower values of @xmath101 that were inferred by @xcite and @xcite . ]
@xcite deduce @xmath108days from the fit to the broad band afterglow light curve , while @xcite infer @xmath109days .
thus we use a value of @xmath110days for grb 020405 . in all cases
the polarization limits are at times either similar to or slightly before the jet break time , @xmath111 .
if the observed radio frequency is below the self absorption frequency , @xmath112 , this can significantly reduce the polarization , for any magnetic field configuration .
@xcite found that the peak of the radio flare emission is typically due to the passage of the self absorption frequency across the observed band .
this would imply @xmath112 and a suppression of the polarization during the rise to the peak , and @xmath113 ( no suppression of the polarization ) during the decay after the peak .
this can reduce the polarization before or around the time of the peak in the radio flare emission . in grbs 991216 and 020405
the radio flux is decaying at the time of the measurement of the radio flare , and direct measurements of the spectral slope also support an optically thin spectrum ( @xmath113 ) , so that no suppression of the polarization is expected due to self absorption . for grb 990123 the point used to measure the polarization seems to be near the peak of the radio flare and there is no good measurement of the spectral slope near this time .
thus the polarization might be somewhat suppressed if @xmath114 , but probably not significantly suppressed since the relatively high flux and the proximity to the peak of the radio flare suggest that @xmath115 is not much smaller than @xmath116 and the two are at most comparable .
since the radio flare emission typically peaks at @xmath111 , the possible lateral spreading of the jet at @xmath117 can be neglected , and the jet dynamics can be reasonably approximated as being part of a spherical flow .
after the passage of the reverse shock , the shocked external medium approaches the ( * ? ? ? * hereafter bm ) self similar solution . however , the shocked ejecta has a significantly higher density than that given by the bm solution , and thus its rest mass density becomes comparable to its internal energy density ( i.e. it becomes cold " ) much earlier on . for a mildly relativistic reverse shock
this happens soon after the reverse shock crosses the shell , while for a relativistic reverse shock the shocked shell is first reasonably described by the bm solution , while it is still hot " , but deviates from that solution once it becomes cold " @xcite . for the bm solution the value of the self
similar variable @xmath118 for a fixed fluid element , that is appropriate for the original ejecta , evolves with radius @xmath64 as @xmath119 for an external density @xmath120 @xcite , where @xmath121 is the radius where it crossed the shock , which for the original shell of ejecta is given by the deceleration radius , @xmath122 .
also , @xmath123 where @xmath124 is the lorentz factor of the fluid just behind the shock , so that @xmath125 for the original ejecta .
more generally one can assume some power law dependence of the lorentz factor on radius , @xmath126 , where the power law index @xmath127 can deviate from its value for the bm solution , @xmath128 .
this implies @xmath129 , @xmath130 and @xmath131 .
once the shell becomes cold " ( i.e. its rest energy exceeds its internal energy ) its dynamics would deviate from the bm solution @xcite , however , the power law index @xmath127 is still bounded by the value just behind the forward shock , @xmath132 ( since the the ejecta shell is lagging behind the shocked external medium ) and by the value for the bm solution , @xmath128 ( since there is a larger inertia that resists the deceleration which is driven by a similar pressure ) .
thus , @xmath133 or @xmath134 for @xmath135 ( a uniform density external medium).a stellar wind environment is not so relevant for the radio flare since in this case the reverse shock electrons are are fast cooling ( i.e. cool significantly due to radiative losses within the dynamical time ) and therefore the observed flux rapidly decays after the passage of the reverse shock , and no detectable radio flare is expected @xcite . ] at any given observed time @xmath136 the observed emission from the reverse shock is from a somewhat smaller radius and a slightly smaller lorentz factor compared to the emission from the forward shock .
this can be seen as follows .
let quantities normalized to their values at @xmath105 be denoted by a twiddle . from the equality of the observed time we have @xmath137 where @xmath138 for the forward shock and @xmath139 for the reverse shock .
since @xmath131 this implies that the ratio of the lorentz factors for a fixed observed time is @xmath140}=\tilde{t}^{-[g-(3-k)/2]/[(4-k)(2g+1)]}$ ] . for @xmath135 this gives @xmath141}$ ] where the exponent ranges between @xmath142 and @xmath143 for the possible range of @xmath134 , which implies @xmath144.s and the radio flare emission is observed at @xmath145s , this implies @xmath146 . ] the somewhat smaller radius of the ejecta shell and the fact that it is typically cold by the time of the radio flare which is typically at @xmath111 suggest that a possible lateral spreading of the jet is probably not important for the radio flare emission , and thus it shall be neglected when calculating the polarization .
if a high polarization in the prompt grb is caused by a viewing angle @xmath13 , then the polarization of the optical flash around its peak would be similar to that of the prompt @xmath0-ray emission .
this is since at this time ( @xmath147 ) @xmath148 and @xmath149 ( i.e. very little lateral expansion could have taken place before the deceleration time ) .
however , at the time of the radio flare ( @xmath66day ) @xmath150 and even a modest lateral expansion would give @xmath151 , so that the jet would occupy our line of sight .
this could significantly reduce the polarization for a magnetic field that is random within the plane of the shock , as is expected to be produced in relativistic collisionless shocks by the two stream instability @xcite .
furthermore , for a shock produced magnetic field the polarization of the radio flare should be similar to that of the optical emission from the forward shock , which is observed to be @xmath152.-ray emission , which requires @xmath153 . since typically @xmath154
, most lines of sight would be inside the jet , @xmath155 . ]
such a low polarization suggests that the magnetic field is not random in 2d , fully within the plane the shock , but is instead more isotropic , and has a comparable component in the direction normal to the shock @xcite .
constraints similar to our best case , grb 991216 , for a larger number of grb , could suggest a similar conclusion for the magnetic field in the ejecta .
our upper limits on the polarization of the radio flare emission can put much tighter constraints on models where there is an ordered magnetic field in the grb ejecta .
if there is a magnetic field in the ejecta that is ordered over patches of angular scale @xmath156 then typically @xmath157 during the prompt @xmath0-ray emission and near the peak of the optical flash , while during the radio flare the polarization can be reduced due to averaging over @xmath158 incoherent patches , @xmath159 $ ] @xcite .
this implies @xmath160 , where @xmath161 is our upper limit on the linear polarization . during the radio flare @xmath162 so that for grb 991216 we have @xmath163 and @xmath164rad .
in particular , if the magnetic field is roughly uniform over the whole jet ( @xmath165 ) then this would imply @xmath157 @xcite , i.e. tens of percent of polarization , which is definitely inconsistent with our upper limits . for a toroidal magnetic field in the ejecta
, we show our upper limits superimposed on the theoretical polarization light curves in fig [ fig1 ] , both for a uniform jet and for a ` structured ' jet .
the latter is expected in models where the grb outflow is poynting flux dominated @xcite .
the polarization light curve for a structured jet is taken from @xcite .
the polarization light curves for a uniform jet are derived in appendix [ app1 ] .
for a uniform jet we consider the two limiting cases for the evolution of the lorentz factor of the ejecta after the deceleration time ( see [ dyn ] for details ) : ( i ) the same as the forward shock lorentz factor , and ( ii ) following the @xcite self similar solution . in case ( ii )
the lorentz factor of the ejecta @xmath0 decreases slightly faster with the observed time @xmath166 ( @xmath167 instead of @xmath168 ) .
this slightly ` contracts ' the polarization light curve along the time axis .
also , while @xmath169 in case ( i ) , we have @xmath170 where typically @xmath171 , which shifts the polarization light curves to earlier times by a factor of @xmath172 .
this can be readily seen by comparing the upper and middle panels in fig .
[ fig1 ] . since @xmath173 @xcite ,
a higher values of @xmath174 produces a higher degree of polarization .
nevertheless , even their lowest values for optically thin synchrotron emission , @xmath175 and @xmath176 , still produce a fairly high degree of polarization . for a uniform jet
, @xmath177 significantly increases with @xmath178 , and goes to zero at @xmath179 .
thus , our upper limits on the polarization put upper limits on @xmath178 .
these limits also depend on the dynamical model for the grb ejecta . for grb 991216
we obtain @xmath180 and @xmath181 for cases ( i ) and ( ii ) , respectively .
the model that is most severely constrained by our upper limits on the polarization is a toroidal magnetic field together with a structured jet ( see lower panel of fig .
[ fig1 ] ) . in this case
all of our upper limits are significantly below the predictions of this model ( by a factor of @xmath182 for grb 991216 ) .
thus a predominantly toroidal magnetic field in the grb ejecta together with a structured jet is hard to reconcile with our upper limits .
the effects of propagation in plasma inside the source on the synchrotron emission in grbs has been recently considered ( * ? ? ? * ; * ? ? ?
* hereafter slw04 ) . in the early emission
while the reverse shock is still going on , these effects can suppress linear polarization and produce up to tens of percent of circular polarization below the self absorption frequency @xmath116 , for a magnetic field that is ordered on large scales ( slw04 ) .
we have obtained upper limits both on the linear polarization ( @xmath183 ) and on the circular polarization ( @xmath184 ) for the radio flare emission .
the best limits are for grb 991216 : @xmath185 and @xmath186 ( @xmath1 ) .
our upper limits directly constrain such plasma propagation effects .
there are some arguments which suggest that plasma propagation effect should be relatively small and sub - dominant in the radio flare emission .
first , the suppression of linear polarization and prevalence of circular polarization is much larger when the reverse shock electrons are fast cooling , i.e. most electrons cool significantly on a time scale smaller than the time it takes the reverse shock to cross the ejecta shell ( swl04 ) , and is less significant when they are slow cooling @xcite . as mentioned in footnote [ sc ] , slow cooling is required in order to have a detectable radio flare emission , and is therefore the relevant case for us .
second , the effects of propagation in plasma decrease with time at a fixed observed frequency @xmath115 @xcite and are significantly smaller at the time of the radio flare ( @xmath187day ) compared to the time when the reverse shock is still going on ( @xmath188s ) .
furthermore , @xmath189 is expected at @xmath113 @xcite , which seems to be the case for the radio flare emission that we consider in this paper .
another related effect that might cause depolarization ( swl04 ) is a different amount of faraday rotation for different frequencies at a given point on the image ( which would be hard to resolve because of the finite instrumental spectral resolution ) or at the same observed frequency between different points on the image ( because of the different path lengths through the emitting plasma and different doppler factor corresponding to different frequencies in the local rest frame of the emitting plasma ) .
the @xmath190 which gives a faraday rotation of @xmath191 for a cold plasma is @xmath192 where @xmath193 is the width of the emitting region in the local frame ( swl04 ) . for a relativistic plasma with a typical electron random lorentz factor of @xmath194 this expression
should be multiplied by @xmath195 ( matsumiya & ioka 2003 ; slw04 ) so that faraday rotation and the resulting depolarization are suppressed .
swl04 consider fiducial parameters that result in fast cooling , and obtain that a significant depolarization is possible up to observed frequencies of @xmath196hz at @xmath197 . since we consider @xmath198hz and faraday rotation scales as @xmath199 , we need to suppress faraday rotation by some 10 orders of magnitude compared to their estimate .
since a detectable radio flare requires slow cooling of the reverse shock , this would imply a smaller density and magnetic field in the local frame , which would reduce the faraday rotation at @xmath197 .
furthermore , slw04 used @xmath200s while the radio flare occurs at @xmath145s .
we have @xmath201 and for the bm solution @xmath202 and @xmath203 so that @xmath204 , which reduces the faraday rotation by @xmath205 orders of magnitude between @xmath105 and the time of the radio flare . also , at the time of the radio flare @xmath206 compared to @xmath207 at @xmath105 so that the same observed frequency corresponds to a higher frequency in the local rest frame of the emitting plasma ( @xmath208 ) .
this reduces faraday rotation by @xmath209 orders of magnitude .
finally , the reverse shock electrons must still be relativistic during the radio flare in order to emit the observed synchrotron radiation , and this will suppress the faraday rotation by a factor of @xmath195 ( swl04 ) .
altogether , it seems that over a large part of the relevant parameter space there will not be significant depolarization of the radio flare emission due to different amounts of faraday rotation at different points on the image , although it is hard to rule out some degree of depolarization for certain regions of parameter space .
if the magnetic field is carried out from the source , then at large radii it would be almost completely in the plane perpendicular to the radial direction , since the radial component of the magnetic field scales as @xmath210 while the two transverse components scale as @xmath211 if the magnetic field is initially tangled on small scales . if the flow and the magnetic field configuration are axially symmetric then @xmath212 while @xmath213 so that @xmath214 will dominate at large radii and the magnetic field will be toroidal .
if the ratio of electromagnetic to kinetic energy is @xmath215 , then the magnetic field is sub - dominant dynamically and the plasma can in principle keep it tangled on small scales within the plane normal to the radial direction , if it is initially tangled on such small scales , and its exact configuration is not obvious from theoretical considerations .
for example , it could be ordered on angular scales @xmath216 where @xmath217 is the half - opening angle of the jet . on the other hand
it might be ordered over the whole jet , possibly in a toroidal configuration . if , on the other hand , the grb outflow is poynting flux dominated ( i.e. @xmath218 ) then the magnetic field is dynamically dominant and can move the plasma around and arrange itself in a toroidal configuration over the whole jet @xcite .
thus an ordered toroidal magnetic field is expected in the latter type of model .
these are the magnetic field configurations that are expected before the prompt gamma - ray emission .
for @xmath215 the gamma - ray emission is likely due to internal shocks , which may cause a shock produced magnetic field that is expected to be random within the plane of the shock , so that it can add a random component to an initially ordered component .
a similar effect can also occur in the reverse shock .
the random component ( @xmath219 ) can reach values close to equipartition , so that it might be dominant over the ordered component ( @xmath220 ) if the latter is well below equipartition ( @xmath221 ) .
however , for @xmath222 with an initially ordered magnetic field , the shock produced random component can at most be comparable to the ordered component . in poynting flux dominated models , where initially @xmath218 , there would be no reverse shock and therefore no radio flare if @xmath223 remains @xmath224 after the prompt gamma - ray emission .
thus , in order to get a radio flare we need @xmath215
. this might still be possible in such a model if after the dissipation of electromagnetic energy ( magnetic reconnection ) that causes the gamma - ray emission , the value of @xmath223 decreases to @xmath225 .
furthermore , the dissipation of magnetic energy ( magnetic reconnection ) that gives rise to the prompt gamma - ray emission in this type of models can also change the local configuration of the magnetic field in the emitting region , so that it might not be perfectly toroidal , and could also have a random component that could potentially be comparable in strength or possibly even exceed the ordered component .
a combination of an ordered and a random magnetic field component could decrease the resulting degree of polarization , compared to that for a purely ordered magnetic field , by a factor of @xmath226 where @xmath227 @xcite . thus we obtain @xmath228 for grb 991216 , where our upper limit on the polarization is @xmath182 times smaller than the predicted value for a purely toroidal magnetic field . in other words , a sufficiently sub - dominant ordered toroidal magnetic field together with a larger random magnetic field component is required .
this requirement is not trivial when starting from an ordered toroidal magnetic field close to the equipartition value , since the shock produced random field component would not exceed equipartition and thus would typically be at most comparable to the ordered component ( @xmath229 ) .
we have derived upper limits on the linear and circular polarization of the radio flare emission discovered from grbs to date .
our results are summarized in table 1 .
there is a reasonably good case that the radio flare emission in grbs 990123 , 991216 , and 020405 indeed arises from the original ejecta that was shocked by the reverse shock and then cooled adiabatically ( as discussed in [ rs ] ) .
this emission is also most likely predominantly synchrotron radiation .
therefore , our upper limits on the polarization can be used to constrain the magnetic field structure in the ejecta of these grbs .
propagation effects outside of the emitting region might decrease the measured polarization below the value of the intrinsic polarization ( see discussion in [ obs ] ) .
we have demonstrated that the relatively small source size , @xmath230 cm , at the time of the radio flare makes it very unlikely that gradients in the rotation measure across the image due to the magnetic field in the ism of the host galaxy would cause significant depolarization .
depolarization due to a possible propagation in a molecular cloud or in the immediate circumburst medium might cause larger gradients in the rotation measure across the image , but still require extreme conditions in order to cause significant depolarization .
propagation effects in the plasma inside the source might cause a different amount of faraday rotation at different points on the image which may cause depolarization since the image is not resolved . for a fixed observed frequency , such a depolarization is much smaller during the radio flare compared to @xmath231 ( see discussion in [ prop_int ] ) .
since faraday rotation scales as @xmath199 it might still be important in the radio for some parameter values , although it should not be very important for a large part of the relevant parameter space .
thus a more thorough investigation of this effect and its magnitude over the relevant parameter space is called for . keeping this caveat in mind
, we continue and examine the implications of our upper limits on the polarization under the assumption that there was no significant depolarization .
we have compared our upper limits to the predictions of different theoretical models .
models where the magnetic field is produced in the shock would in most cases produce a polarization below our upper limits .
furthermore , they are expected to produce a polarization similar to that of the afterglow emission , if indeed the dominant magnetic field in the afterglow is shock produced and if the magnetic field configuration behind the afterglow shock is similar to that behind the reverse shock . for lines of sight near the edge of the jet the expected polarization might in some cases exceed our upper limits , if the magnetic field behind the shock is maximally anisotropic ( i.e. random within the plane of the shock or ordered in the direction normal to the shock ) .
this might suggest a more isotropic magnetic field configuration behind the shock if comparable or better upper limits will be found in a larger sample of grbs , similar to the current situation with the afterglow emission where linear polarization has been measured in several grbs and was found to be @xmath232 in all cases , perhaps with one exception
grb 020405 where a sharp spike in the polarization ( @xmath233 at @xmath234days ) was reported @xcite .
we note that an almost simultaneous polarization measurement by a different group @xcite resulted in a significantly lower polarization , @xmath235 .
our upper limits on the polarization put much stronger constraints on models in which there is an ordered magnetic field in the ejecta . a magnetic field
that is roughly uniform across the whole jet would produce tens of percent of polarization , which is inconsistent with our upper limits .
if the magnetic field is ordered over patches of angular scale @xmath236 , which are mutually incoherent , then our upper limits on the polarization put upper limits on @xmath236 .
the tightest constraint is for grb 9901216 , for which we find @xmath237rad ( see [ imp ] ) .
the polarization light curves for a toroidal magnetic field in the ejecta , together with our upper limits , are shown in fig .
[ fig1 ] . for
a uniform jet , our upper limits on the polarization constrain our viewing angle toward grb 991216 to be @xmath238 where this range roughly covers the uncertainty in the dynamics of the ejecta at @xmath136 ( as discussed in [ imp ] ) .
these values are for a conservative value of the spectral slope , @xmath175 which corresponds to @xmath176 . for a structured jet with a toroidal magnetic field , @xmath157 is expected at @xmath239 ( which applies for all the upper limits given in table 1 ) . therefore , this model appears to be inconsistent with our upper limits on the polarization .
we thank ehud nakar , tsvi piran , davide lazzati and elena rossi for useful discussions .
this research was supported by us department of energy under contract number de - ac03 - 76sf00515 ( j.g . ) .
the national radio astronomy observatory is operated by associated universities , inc .
, under cooperative agreement with the national science foundation .
here we derive the polarization of a uniform jet with an ordered toroidal magnetic field , by generalizing the results of @xcite for a uniform transverse magnetic field .
the emission is assumed to arise from a section of a thin spherical shell moving radially outward with a bulk lorentz factor @xmath240 , that lies within a cone of half opening angle @xmath217 , which represents the jet . for simplicity ,
the emission is integrated over the jet at a fixed radius and the the differences in photon arrival time from different angles @xmath241 from the line of sight are ignored .
an integration over the equal arrival time surface of photons to the observer might introduce small quantitative differences , but the results should be qualitatively similar , as we verified by comparing our results to those of @xcite . the emission in the local rest frame of the shell ( where quantities are denoted by a prime ) is taken to be uniform across the jet ( hence a uniform jet ) , and depends only on the angle @xmath242 between the direction of the emitted radiation , @xmath243 , and the local direction of the magnetic field , @xmath244 . the polarization position angle at a given point on the jet
makes an angle of @xmath245\cot\phi\}$ ] from the local direction of the magnetic field , @xmath246 @xcite , where @xmath247 , and @xmath248 is the angle between @xmath246 and the direction from the line of sight to that point on the jet .
the stokes parameters are given by @xmath249 , where @xmath6 is measured from some fixed direction , which for convenience we choose to be the direction from the jet symmetry axis to the line of sight .
we have @xmath250 where @xmath251 is the azimuthal angle around the line of sight measured from the direction between the jet axis and the line of sight .
let us use the notations @xmath252 , @xmath253 , @xmath254 , and @xmath255 .
we have @xmath256 with @xmath257 , @xmath258 and @xmath259 where @xmath243 is the direction in the local frame of a photon that reaches the observer .
we also find that @xmath260\ .\ ] ] from symmetry considerations @xmath261 and @xmath262 .
the direction of the polarization vector on the plane of the sky is along the line connecting the jet symmetry axis and the line of sight .
the degree of polarization is given by @xmath263\ ] ] @xmath264^{-1}\ , \ ] ] where @xmath265 is the heaviside step function , @xmath266/[2q(y_jy)^{1/2}]$ ] , @xmath267 , and for producing the results shown in fig .
[ fig1 ] we use @xmath268 @xcite . in order to produce polarization light curves a simple model of @xmath269
is added , where @xmath166 is the observed time . for a jet with no lateral
spreading going into a uniform density medium @xmath270 where @xmath271 $ ] and @xmath126 ( see [ dyn ] ) .
we identify the jet break in the optical light curve with @xmath272 where @xmath273 is the lorentz factor just behind the forward shock .
thus @xmath274/[(4-k)(2g+1 ) ] } \left(\frac{t}{t_j}\right)^{-g/(2g+1)}\ .\ ] ] as discussed in [ dyn ]
we have @xmath133 . for the lower limit @xmath132
which corresponds to @xmath275 we have @xmath276 . for the upper limit @xmath128
we have @xmath277}(t / t_j)^{-(7 - 2k)/[2(4-k)]}$ ] .
|
the magnetic field structure in gamma - ray burst ( grb ) outflows is of great interest as it can provide valuable clues that might help pin down the mechanism responsible for the acceleration and collimation of grb jets .
the most promising way of probing this magnetic field structure is through polarization measurements of the synchrotron emission from the grb ejecta , which includes the prompt @xmath0-ray emission and the emission from the reverse shock .
measuring polarization in @xmath0-rays with current instruments is extremely difficult : so far there is only one claim of detection ( a very high degree of linear polarization in grb 021206 ) which despite the favorable conditions remains highly controversial .
the emission from the reverse shock that propagates into the ejecta as it is decelerated by the ambient medium peaks in the optical on a time scale of tens of seconds ( the so called ` optical flash ' ) and dominates the optical emission up to about ten minutes after the grb .
unfortunately , no polarization measurements of this optical emission have been made to date . however , after the reverse shock finishes crossing the shell of grb ejecta , the shocked ejecta cools adiabatically and radiates at lower and lower frequencies .
this emission peaks in the radio after about one day , and is called the ` radio flare ' .
we use vla data of radio flares from grbs to constrain the polarization of this emission .
we find only upper limits for both linear and circular polarization .
our best limits are for grb 991216 , for which we find @xmath1 upper limits on the linear and circular polarization of @xmath2 and @xmath3 , respectively .
these limits provide interesting constraints on existing grb models .
specifically , our results are hard to reconcile with a predominantly ordered toroidal magnetic field in the grb outflow together with a ` structured ' jet , where the energy per solid angle drops as the inverse square of the angle from the jet axis , that is expected in models where the outflow is poynting flux dominated .
|
many experiments have attempted to directly detect weakly interacting massive particle ( wimp ) dark matter candidates via their elastic scattering on target nuclei @xcite .
the lux experiment has recently limited the spin - independent cross sections to be under @xmath3 for 33 gev wimp mass @xcite .
a method that has commonly been used is to measure signals associated with the deposited energy of the nuclear recoil .
however , background processes , such as those induced by neutrons , can mimic wimp signals @xcite . to overcome this problem , a
`` smoking gun '' wimp signal would be its unique directional event - rate dependence @xcite . when viewed from the earth ,
the average wimp velocity in galactic coordinates is from the cygnus direction , while the directions of background sources are fixed in an earth - based coordinate system .
this effect can be seen by the directions of the recoiling nuclei that are strongly correlated with the directions of the incoming wimps @xcite , and produce a diurnal variation of rates due the earth s rotation @xcite .
prospects for a working directional detector have focused on low pressure gas time projection chambers , such as drift - ii @xcite , dmtpc @xcite , newage @xcite , mimac @xcite , and the d@xmath4 prototype @xcite , and emulsion techniques @xcite .
the sensitivity of those detectors have been studied assuming they are either capable of reconstructing full three - dimensional ( 3d ) tracks @xcite or only tracks projections on a particular detector - fixed plane ( 2d detector ) @xcite .
it is found that of order tens of events are required to reject isotropy of recoil angle distributions , and the number of required events are one order of magnitude larger for a detector that can not measure the sense , defined as the absolute sign ( @xmath5 or @xmath6 ) , of the recoil vector .
however , it is still a daunting task to fully reconstruct sub-100 kev nuclear recoil tracks . in this paper
, we consider a detector that can detect the polar angle of the recoil track with respect to a fixed @xmath7 axis , while the azimuthal angle is not measured . in the following ,
we refer to this as a polar detector . in general ,
since reconstruction of recoiling nucleus tracks is not required for a polar detector , the experimental realization can be more reliable and feasible .
one issue for a directional dark matter detector is whether or not it possesses sense recognition capability , where we call the former a head - tail detector and the latter an axial detector .
considerable technological effort is required to provide head - tail detection capability . throughout the paper ,
the standard isothermal galactic halo model is used to model the wimp velocity distribution . in this case
, a head - tail polar detector with @xmath7 axis aligned with the wimp wind direction has the same performance as a full 3d detector , because in this orientation the azimuthal distribution is completely flat and provides no information at all .
one example of an axial polar detector is a stilbene crystal , which is an organic single crystal whose scintillation efficiency depends on the nuclear recoil direction relative to crystallographic axes @xcite .
recently , a new technique of columnar recombination ( cr ) has been shown to be capable of realizing a working axial polar detector @xcite , without sense detection capability .
the directional sensitivity in a cr detector comes from the dependence of electron - ion recombination level on the angle between recoil track and electric field . here
the experimental signals are the scintillation light for the recombinations , and drift electrons from the surviving ionizations . in most previous sensitivity studies such as those reported in refs .
@xcite , the recoil energy is integrated out and only distributions with respect to angular variables are studied .
when all the energies are integrated out , the shape of angular distribution does not depend on the wimp mass @xcite and , thus , the distribution and the analysis are greatly simplified .
however , we note that in a real experiment , all relevant information are used in order to achieve the highest possible sensitivity .
thus , both energy and directional information are used simultaneously in this work .
previous studies that did use both the energy and angular information @xcite , treated the wimp mass as a known parameter that was kept at a fixed value in the fit . while this procedure is simpler to implement for producing plots of cross - section upper limits versus wimp mass , it is problematic for a general directional sensitivity calculation . in this study
, we address this issue by treating the wimp mass as a nuisance parameter in a statistical framework that uses the well - established profile likelihood method .
we study signals from spin - independent wimp - nucleus interactions , and assume zero background to provide benchmark results . for nonzero background experiments ,
an observation of directionality still unambiguously leads to an observation of wimps , and the amount of directional sensitivity loss due to background is also studied . in sections that follow ,
we will start by calculating the distribution of observables , and then find the best orientations for various detector configurations using the standard profile likelihood method .
later , the effects of background contamination are calculated for detectors in optimal orientations .
finally , we study the performance of detectors using xenon target with various energy thresholds and wimp masses , and estimate the required detector exposure to see a three sigma directional signal .
the general double - differential derivative of the recoil rate per unit target mass , with respect to nuclear recoil energy @xmath8 and solid angle @xmath9 in the nuclear recoil direction @xmath10 , can be expressed as @xcite : @xmath11 here @xmath12 is the reduced mass of the wimp - nucleus system , where @xmath13 and @xmath14 are the wimp and target nucleus mass , respectively .
@xmath15 is the form factor , @xmath16 is spin - independent wimp - nucleus cross - section , and @xmath17 is the number density of the wimp , where we use @xmath18 .
the minimum wimp speed required for a recoil of energy @xmath8 is @xmath19 .
the radon transform , @xmath20 , of the wimp velocity distribution @xmath21 represents the sum of the probability densities where the velocity projection on direction @xmath10 is equal to @xmath19 : @xmath22 in the standard isothermal galactic halo model , the distribution of wimp velocities @xmath23 relative to the target is maxwellian , with an average value of @xmath24 : @xmath25 in this case the radon transform is : @xmath26 the maximum recoil energy for a wimp with velocity @xmath27 is @xmath28 , and by reversing the formula , we obtain @xmath29 , with the definition @xmath30 .
we also calculate the standard total rate @xmath31 when @xmath32 to be @xmath33 . for a a single target - nucleus type , the differential rate exhibits more physical meaning
when it is expressed using @xmath34 and @xmath31 : @xmath35 it is convenient to define two variables : @xmath36 in a reference frame where the reference @xmath7 axis is parallel to the wimp wind direction , or the cygnus direction , the solid angle dependence reduces to polar angle @xmath37 dependence , and the differential rate can be expressed as : @xmath38 for simplicity , the peculiar velocity of the sun and the earth s orbital velocity about the sun are ignored and the velocity of earth @xmath39 is taken to be equal to the velocity of the local standard of rest @xmath40 . in this case the magnitude of @xmath39 and @xmath41 is same , which is usually given the value of 220 km / s .
hence in this paper we use @xmath42 to represent the standard halo model .
it is feasible to build a detector that is oriented at a fixed direction in space , where the wimp wind direction and the @xmath7 axis of detector form a fixed angle @xmath43 . in the frame of a detector system
, the polar and azimuth angles of recoil @xmath44 and @xmath45 provide the directional information and the @xmath46 term in eq .
( [ eqn : drdedcos ] ) can be expressed in terms of @xmath44 and @xmath45 as @xmath47 . here
the wimp wind direction is assumed to lie in the @xmath48 plane , without any loss in generality .
the resulting differential rate in lab frame with respect to @xmath8 , @xmath49 and @xmath45 is then : @xmath50 here , a solid angle transform @xmath51 is applied from eq .
( [ eqn : drdedcos ] ) . for a detector where the only directional measurable is the polar angle @xmath44 , the azimuthal angle @xmath45 should be integrated out : @xmath52 for an axial detector where the sense of the recoil direction can not be distinguished
, the distributions should be folded , as @xmath53 , and @xmath54 .
now we consider a common case where a detector is fixed in the earth s coordinate frame .
since the earth itself is rotating , the angle @xmath43 modulates with a period of one sidereal day .
the wimp wind from cygnus is directed to the earth at a constant angle @xmath55 to the earth s polar axis , as shown in fig .
[ fig : earth - dt ] .
if the @xmath7 axis of the detector is oriented relative to earth s polar axis at a fixed angle @xmath56 , then the dependence of the angle @xmath43 as a function of time @xmath57 in units of sidereal days is : @xmath58 here @xmath57 is zero when the wimp velocity lies in the plane made by the detector @xmath7 axis and earth s polar axis .
since @xmath57 distributes uniformly within 0 and 1 when all energies and angles are considered , the appropriate distribution for the earth - fixed case , now in terms of ( @xmath59 ) , @xmath60 , is determined by replacing @xmath43 in eq .
( [ eqn : drdedcosldphil ] ) with eq .
( [ eqn : costheta0 ] ) and integrating out @xmath45 .
note that here , instead of using angular information only as in @xcite , we use two and three dimensional distributions that include energy and sidereal time as additional variables .
the most common method that is used in particle physics to determine significance of an observation is the profile likelihood ratio test statistic method .
it uses a hypothesis test against a null or trivial hypothesis @xmath61 , where the data is assumed to correspond to the distribution of an alternative or interesting hypothesis @xmath62 .
usually , a set of parameters of interest @xmath63 of @xmath62 are fixed to a specific value @xmath64 to obtain @xmath61 .
a likelihood ratio is calculated as the maximum likelihood of a null hypothesis divided by the maximum likelihood of an alternative hypothesis , with nuisance parameters @xmath65 floating : @xmath66 the test statistic @xmath67 is defined as @xmath68 .
the @xmath69 value is the probability to have a discrepancy larger than the observed one @xmath70 , as @xmath71 the smaller the @xmath69 value , the more credible that hypothesis @xmath61 is not correct .
for example , a @xmath69 value of 0.00135 corresponds to a @xmath72 signal significance . in principle , the distribution @xmath73 needs to be obtained using monte carlo methods with high statistics .
however , according to wilk s theorem , we can assume @xmath67 follows a @xmath74 distribution @xcite for data sets in our study
. then the @xmath69 value can be directly calculated as the probability above @xmath70 for a @xmath74 distribution @xmath75 with degrees of freedom equal to the number of fixed parameters in the numerator of eq .
( [ eqn : lratio ] ) : @xmath76 .
the significance @xmath77 is related with @xmath69 via @xmath78 , where @xmath79 is the cumulative function of the standard gaussian .
the likelihood function @xmath80 used in this paper is the unbinned product of normalized probability density function ( pdf ) of observed quantities for all events .
the pdf is normalized over two or three dimensional variables , such as @xmath81 or with the inclusion of @xmath57 .
a nonzero @xmath82 indicates a finite average wimp speed from the cygnus direction .
thus , for studies in this paper , @xmath82 as @xmath63 is the sole parameter of interest , with the null hypothesis being @xmath83 .
the wimp mass , since it is unknown , is a nuisance parameter and floated in the fits for both the numerator and denominator of eq .
( [ eqn : lratio ] ) . unlike most of the studies such as @xcite , where a fixed wimp mass need to be inserted in order to get the sensitivity or limit , our method directly focuses on the directionality itself . to test the sensitivity of a measurement , the median ( not mean ) value of the test statistic @xmath84 of a large ensemble of hypothesis tests
is used as a measurement quantity of sensitivity for the work reported here .
the ensemble size @xmath85 is 1000 for all of calculations discussed below .
the error of @xmath84 is @xmath86 , where @xmath87 is the normalized probability density function that is obtained from the raw distribution of @xmath67 for the ensemble by a kernel density estimation method @xcite . here
, for each hypothesis test , a toy monte carlo data set with a certain number of events is generated and fitted to obtain the @xmath67 value for this data set . with only one parameter of interest , the significance in units of @xmath88
is calculated as the square root of the test statistic : @xmath89 .
so a @xmath90 significance corresponds to @xmath91 .
in all the studies of this work , @xmath84 is found to be proportional to the total event number , as will be shown in fig .
[ fig : ne_q - n10 ] .
so we use a linear proportional function to fit @xmath84 , and find the number of events corresponding to @xmath92 as the required size for a @xmath72 discovery . because of the linearity , the sensitivity can also be scaled proportionally to the size of statistics when needed .
values on solid markers as a function of cosine of the angle between the detector @xmath7 axis and the wimp wind direction for general space - fixed detectors .
head - tail polar and angular only detectors correspond to up and down triangles .
axial polar and angular only detectors correspond to circles and diamonds .
the event number for head - tail configuration is 200 , different from 50 events for axial configuration , in order to fit in a single plot . ]
values on open markers as a function of angle @xmath56 between detector @xmath7 axis and earth s pole , for earth - fixed detectors .
general head - tail polar and angular only detectors with 50 events correspond to up and down triangles .
general axial polar and angular only detectors with 200 events correspond to circles and diamonds . for a standard axial xenon detector of 3 kev threshold with 200 observed events , four graphs for wimp masses of 20 , 30 , 50 and 100 gev
are shown as blue squares with decreasing sizes .
these graphs are fitted using an empirical function @xmath93 that are superimposed as red solid curves for a standard axial xenon detector , and as red dashed and dotted curves for general axial polar and angular only detectors . ]
values versus numbers of observed events for all detector types in their optimal orientations .
solid squares represent a standard space - fixed axial xenon detector .
other symbol markers are explained in figs .
[ fig : q_vc - sf4 ] and [ fig : q_vc - rt5 ] .
the fitted linearly proportional lines are superimposed .
blue dashed lines represent axial xenon detectors configured at 30 gev wimp and 3 kev threshold . ]
to study the general performance of a polar detector , a general detector with zero energy threshold and unit form factor is considered . in the absence of the form factor term in eq .
( [ eqn : drdedcosldphil ] ) , the energy dependence of the rate is expressed through the ratio @xmath94 , i.e. , in units of @xmath34 .
since the wimp mass dependence of the rate function only enters through @xmath34 , a change of wimp mass just changes @xmath34 , the overall scale of the energy @xmath8 .
thus the sensitivity for the general polar detector has no dependence on the wimp mass . in secs .
[ sec : gen][sec : bkg ] , general results applicable to all wimp masses are shown . figure [ fig : q_vc - sf4 ] shows the median value of @xmath67 for space - fixed detectors , with head - tail and axial configurations , for different detector orientation angles .
we can see that the optimum orientation for a space - fixed case is always at @xmath95 , where the detector s @xmath7 axis points toward the cygnus .
this is expected since in this case the detector can gain maximum information from the measured @xmath49 .
nontrivial curves for axial detectors are discovered with a minimum at around @xmath96 , where the distribution of @xmath97 is close to flat due to the folding of positive and negative @xmath49 values . in the earth - fixed case ,
the quantity @xmath84 is averaged over many directions , and depends on the orientation @xmath56 .
figure [ fig : q_vc - rt5 ] shows the dependence of @xmath84 on the angle @xmath56 .
a general head - tail detector is shown to have the most sensitivity when the @xmath7 axis is oriented parallel to the earth s polar axis ( @xmath98 ) .
interestingly , the optimal @xmath56 value for a general axial detector is near @xmath99 .
a fit using an empirical function @xmath93 gives the optimum value @xmath100 , with the fitted curve shown in fig .
[ fig : q_vc - rt5 ] as a dashed red line .
the reason for @xmath101 being slightly larger than 42 degree , is that it is more advantageous to lie inside than outside for the wimp wind vector , with respect to the cone made by the detector s @xmath7 axis following the earth s self - rotation .
note that this value @xmath101 will change if the angle of the wimp wind to the earth s pole is a value different than 42 degree .
results for the required number of events for @xmath72 discovery for four detector types , i.e. , @xmath95 for space - fixed detectors , and @xmath102 for earth - fixed axial ( head - tail ) detectors , are displayed in row one of table [ tab : ensxe_neq9_n12 ] .
of order 10 to 20 event are required for head - tail detectors .
these results show that a change from a space- to earth - fixed basis worsens the sensitivity by factors of @xmath103 and @xmath104 for axial and head - tail configurations . for an earth - fixed detector ,
we conclude that the loss of head - tail capability causes the sensitivity to be a factor of 10 worse , and the axial one is less efficient than a full 3d one ( column 2 ) by about a factor of 19 .
these numbers can be compared to the 1d readout in ref .
@xcite , which gives values of 8 and @xmath105 for the two factors , by measuring the signal yield with a constrained background . as discussed in the introduction
, the optimal space - fixed head - tail detector behaves the same as a full 3d detector with its @xmath7 axis aligned along the wimp wind . by dividing the two numbers for the space - fixed case , we expect that an axial detector would have the same sensitivity as a full 3d detector , provided it accumulated @xmath106 times as many events .
this is a much better conclusion than that in ref .
@xcite , which did not study a space - fixed detector .
0.9l|>x|>x|>x|>x & & + & axial & head - tail & axial & head - tail + general & @xmath107 & @xmath108 & @xmath109 & @xmath110 + angular only & @xmath111 & @xmath112 & @xmath113 & @xmath114 + xenon , 30 gev wimp , 3 kev threshold & @xmath115 & & @xmath116 & + xenon , 30 gev wimp , 10 kev threshold & @xmath117 & & @xmath118 & + xenon , 50 gev wimp , 30 kev threshold & @xmath119 & & @xmath120 & +
most previous estimates of the sensitivity of a directional dark matter detector have only considered angular information only . to study detectors with polar angle information only
, the rate as a function of @xmath49 is obtained from eq .
( [ eqn : drdedcosl ] ) by integrating out the energy part : @xmath121 the same statistical procedure is performed as for the general polar detector case , for space- or earth - fixed , axial or head - tail configurations .
the @xmath84 dependence on detector orientation is shown in figs .
[ fig : q_vc - sf4 ] and [ fig : q_vc - rt5 ] . in all four detector configurations ,
the shape of the @xmath122 and @xmath56 dependence is similar to that for the general polar detector case . for a space - fixed axial detector ,
the dependence on @xmath122 is also not monotonic . in the earth - fixed case , the red dotted curve in fig .
[ fig : q_vc - rt5 ] has a maximum at @xmath123 .
results for the optimal value , @xmath95 for space - fixed , @xmath99 and @xmath124 for earth - fixed axial and head - tail detectors , are reported in row 2 of table [ tab : ensxe_neq9_n12 ] . for head - tail detectors , of order of 10 ( 20 ) events
are required to see directionality in space ( earth)-fixed basis , similar to results from previous studies @xcite for 3d and 2d detectors .
the sensitivity worsens by factors of @xmath125 and @xmath104 for axial and head - tail configurations , respectively , when changing a space - fixed to an earth - fixed basis , similar to the general polar detector case . by comparing the event numbers in columns 1 and 3 of table [ tab : ensxe_neq9_n12 ] , i.e. , for the axial configurations of general and angular only detectors , we can see that @xmath126 and @xmath127 larger statistical samples are needed when the energy information is not used . however , for head - tail detectors ( columns 2 and 4 ) , only @xmath128 and @xmath129 times the statistics are required .
this is because in the absence of the powerful head - tail information , which directly relates to directionality , we need to rely on the correlation between recoil energy and direction to improve the directional sensitivity .
thus , the importance of combining energy and angular observables is established , at least for axial detectors . we note that the performance may be improved over that reported in refs .
@xcite , if the observed energy is simultaneously used .
in nondirectional dark matter searches , the most dangerous backgrounds are those with the same energy - dependent shape as the signal , since in those cases one can not distinguish the background by the energy distribution . in this study
, we use the directional observables to distinguish the signal , while the directional information is diluted by the background contribution . consider a background that is distributed the same way as the wimp signal , while not requiring the relative speed between the earth and the dark matter halo . by setting @xmath83 and a unit form factor in eq .
( [ eqn : drdedcosldphil ] ) , the background rate @xmath130 is isotropically distributed with a simple exponential decaying energy dependence : @xmath131 . in a real experiment , usually we do not fully understand all the sources of background . here
we show that this is not a problem in our method . using the same pdf [ eq .
( [ eqn : drdedcosl ] ) ] to fit data , regardless of the how the data is generated or originated as would be in a real experiment , a nonzero @xmath82 still unambiguously indicates a directional signal .
we use the same statistical procedure , and only change the generation of the signal - only data set to a mixture of signal and background .
the parameter of interest is still @xmath82 .
our method avoids the need for an external constraint to be imposed on the background rate , as done , for example , in ref .
@xcite to obtain a better control of the signal component . the sensitivity dependence on the signal purity is shown in fig .
[ fig : q_fs - sb4 ] , where the signal component yield is a constant for each detector type .
all polar detectors are aligned in their optimal orientations .
the @xmath132-axis is the signal purity , defined as the ratio of the input signal number over total event number in the generation . in fig .
[ fig : q_fs - sb4 ] , the sensitivity level can be seen to be almost linear with the signal purity .
as the signal purity goes lower , the relative sensitivity falls by a lesser amount , especially for the space - fixed axial case .
thus we conclude the polar detector has a guaranteed performance level in the presence of background .
in particular , the required number of events for purity 0.4 only increases by about a factor of 2 compared to the pure signal case .
this can be compared to the 1d case of fig . 2 in ref .
@xcite , in which case the degradation of sensitivity is significantly worse as the signal purity decreases .
note that , other types of backgrounds , such those with a flat energy dependence , are easier to handle because the difference in the energy distribution gives extra information to distinguish them from the real signal .
now we consider a detector with a specific material and a minimum energy threshold .
since the proposed columnar recombination detector uses gaseous xenon as target , a xenon helm form factor @xcite @xmath133 is computed and inserted into eq .
( [ eqn : drdedcosldphil ] ) . for the minimum energy threshold ,
the lux experiment @xcite provides a good reference , where the trigger efficiency reaches 50% for 4 kev nuclear recoil events . in gaseous xenon
, the commonly used reflective material polytetrafluoroethylene ( ptfe ) has a low refractive coefficient of about 60% for the 175 nm scintillation wavelength of pure xenon , compared to more than 90% @xcite in liquid xenon .
an experimental prototype @xcite using pure xenon has put a high threshold of 30 kev , mainly due to a low light collection efficiency of 3% .
a small amount of trimethylamine ( tma ) mixture in the xenon gas , can convert the primary excitation to ionization of tma molecules through the penning transfer process @xcite , and increases the cr effect since more ionizations will participate in the recombination process @xcite . at the same time
, the scintillation light is wavelength - shifted to the tma emission band centered at near - uv ( 300 nm ) , where the reflectance of ptfe increases to 90% @xcite .
thus , it should be possible to achieve an energy threshold much lower than 30 kev . in this study
, we consider three minimum energy threshold values of 3 , 10 and 30 kev . because of the rapidly falling shape of the form factor
, the event rate will be suppressed by a factor more than @xmath134 for recoil energies greater than 100 kev .
thus , the maximum of the detection energy range is chosen to be 100 kev .
this standard xenon detector can be viewed as a columnar recombination detector with perfect resolution .
so we only study the axial configuration .
first , in the earth - fixed case , the sensitivity dependence on @xmath56 is studied for four typical wimp mass values of 20 , 30 , 50 and 100 gev .
figure [ fig : q_vc - rt5 ] shows the results for a 3 kev threshold , with fitted curves superimposed .
the fitted optimal values for @xmath135 are @xmath136 degree for 20 , 30 , 50 and 100 gev wimp mass . these are all nearly equal ; a weighted average value is @xmath137 . in the following ,
the optimal directions @xmath99 and @xmath95 are used for earth- and space - fixed standard xenon detectors .
because of the detection threshold , the required number of observed events for detectors with different thresholds can not be directly translated to detector performance , since in this case the total rate depends on the threshold .
the number of observed events , @xmath138 , is the integral of the differential rate within the detection range including the form factor for a given total target mass @xmath139 and measurement time @xmath140 : @xmath141 to make fair comparisons of the sensitivities , we introduce a quantity called the number of point interactions , @xmath142 , which is the number of wimp - nucleus interactions with no threshold and no form factor effects : @xmath143 the amount of required @xmath142 directly relates to detector performance . for general detectors discussed in previous sections , @xmath144 . for a single target
, the spin - independent wimp - nucleus cross - section @xmath16 is related to the wimp - nucleon cross - section @xmath145 as @xmath146 , where @xmath147 is the wimp - nucleon reduced mass .
for xenon , @xmath148 . values as a function of wimp mass for axial xenon detectors , normalized to 200 events of point interaction . from the top to bottom , the dashed , dotted and dot - dashed lines corresponds to the 3 , 10 and 30 kev energy threshold .
the best orientations on space - fixed ( @xmath95 , main plot ) and earth - fixed ( @xmath99 , inset plot ) basis are used . ]
there will be some wimp - mass dependence of the sensitivity since the form factor and the threshold depends on absolute energy values that do not scale with @xmath34 .
the wimp mass is still a floating nuisance parameter in the fitting , while being fixed at various input values in event generation .
figure [ fig : q_mw - xffc2-mw2 ] shows the raw sensitivity as a function of wimp mass , for space- and earth - fixed detectors at their optimal orientations .
each point in the plot is normalized to @xmath149 , so that the sensitivities can be compared for the same detector exposure . in general , the higher the threshold , the lower the sensitivity .
the decrease of the sensitivity for low wimp masses is due to the effect of the threshold that eliminates a higher fraction of events the lower the wimp mass .
the sensitivity decrease for high wimp masses is due to the form factor , which suppresses the high energy recoils .
the most sensitive wimp - mass value increases as the energy threshold increases , and is around 30 gev for a 10 kev threshold and 50 gev for a 30 kev threshold .
the inset of fig .
[ fig : q_mw - xffc2-mw2 ] shows that the performance of earth - fixed detectors follows the same behavior as space - fixed case with respect to wimp - mass and energy - threshold dependence , except for an overall sensitivity that is lower by a factor of about three , which is consistent with the general study in sec .
[ sec : gen ] .
we can also compare the decrease in sensitivity to the decrease in the observed event rate due to the energy threshold cut .
for example , for a 30 gev wimp , an energy threshold change from 3 kev to 10 kev , decreases the sensitivity by only 23% , compared to the 71% decrease in the total number of observed events .
this is due to the fact that most directional information comes from the high recoil energy events , so that a cut on low energy events has minor impact .
for a detector with high energy threshold of 30 kev , a wimp mass in the intermediate range around 50 gev is most promising , with the sensitivity that is 3.2 times below that of a 3 kev - threshold detector for the same range .
last three rows of table [ tab : ensxe_neq9_n12 ] show the required point interaction numbers for a @xmath72 discovery under several wimp - mass and energy - threshold combinations . in a typical case of a 30 gev wimp mass and a 10 kev energy threshold ,
the numbers are @xmath150 and @xmath151 , for space- and earth - fixed xenon polar detectors .
these correspond to @xmath1 and @xmath152 kg@xmath2year exposures for a @xmath153 spin - independent wimp - nucleon cross - section .
a columnar recombination detector using high pressure xenon - tma gas mixture can measure the angle between the nuclear recoil track and electric drift field @xcite , without sense recognition capability .
thus it is an axial polar detector .
the columnar effect benefits from a high penning efficiency as the percentage of the xenon excitations that fall back to free electrons by ionizing the tma molecules . for electron and gamma energy deposit ,
the penning efficiency is estimated to be around 10% and 20% in ref .
@xcite and ref .
@xcite , and also indirectly measured to be 10%-15% in high pressure xenon gas with tma mixture @xcite .
in addition , the primary scintillation from xenon is observed to be absorbed by the tma mixture @xcite .
nevertheless , the columnar recombination effect for the ionization channel has been seen for the alpha particle events as shown in fig . 5 of ref .
@xcite , where , as a first approximation , the collected charge is a linear function of @xmath154 .
further studies are needed to obtain a high penning efficiency for the high initial ionization density produced by nuclear recoils . to date
, no experimental results on the properties of nuclear recoils in a columnar recombination material is available .
once a reliable detector response is available , a detector resolution can be straightforwardly included in our model to obtain the sensitivity . despite this limitation
, we have tested a naive resolution smearing method for nuclear recoils based on the following considerations : ( a ) the recombination fraction of the electron - ion pairs ranges from 0.8 for recoils that are parallel(@xmath155 ) to 0.4 for those that are perpendicular ( @xmath156 ) to the field direction ; ( b ) an ionization work function of 76 ev for the electron - ion pairs ; and ( c ) 10% and 50% efficiency for the recombination ( scintillation ) and ionization channels with gaussian resolutions calculated from a fano factor @xcite of 0.14 @xcite for xenon gas .
the resulting sensitivity is only worsened by 7% relative to the zero resolution case .
this can be understood as that the directionality is derived from an overall anisotropic phenomenon and not from a localized spot .
reference @xcite also noted that variations in angular resolution do not make much difference in the dark matter detection sensitivity .
thus we do not expect a significant change to our sensitivity results by detector resolution effects .
the performance of a directional dark matter detector with polar angle detection is studied for various configurations .
a wimp - mass independent method is used to obtain the sensitivity of a general detector .
in addition , a detector with xenon as target material is studied .
we infer that : * both axial and head - tail polar detectors have the highest sensitivity when the @xmath7 axis is aligned with wimp wind .
however , the dependence of sensitivity to detector orientation is not monotonic for an axial detector . to obtain optimal performance when rotating with the earth
, the @xmath7 axis should be oriented at 45 degree to earth s pole for an axial detector , while it should be aligned with earth s polar axis for a head - tail detector .
* a head - tail polar detector can detect directionality with of order 10 or 20 events on a space- or earth - fixed basis . in the absence of sense detection capability , an order of magnitude more statistics is needed . * without using energy information simultaneously , the required statistics would be a factor of @xmath157 times higher for a space ( earth)-fixed axial detector
. this conclusion will be useful for detector types in which partial directional information is available , such as a 2d planar detector . * a general axial polar detector with @xmath106 times the statistics has the same performance as a general full 3d tracking detector .
however , in experimental practice , the target mass for a full 3d detector is limited because of diffusion effects , and accomplishing millimeter tracking is extremely challenging . on the other hand
, a detector with polar angle sensitivity without head - tail discrimination requirement can use straightforward experimental techniques .
in addition , it can be made in large volume with high - density gas , with a target mass that can be orders of magnitude larger than a conventional full 3d detector .
we conclude that it is of great advantage to explore the directional dark matter detection technique using a senseless polar angle detection apparatus .
* a space - fixed detector is generally found to be 3 and 2 times more sensitive than an earth - fixed detector , for axial and head - tail configurations .
this ratio is an important factor when comparing the additional cost for a space fixed detector , since it has to rotate all the time with respect to the earth . in a space - fixed detector , to distinguish between the wimp signal from the galaxy coordinate and possible anisotropic background originating from the detector frame , manually reversing the detector @xmath7 axis direction for half of the measurement period should be useful .
* in the presence of contamination by an isotropic background that mimics the energy shape of the signal , the decrease of directional sensitivity occurs at a rate that is less than the decrease of signal purity .
thus a polar detector is robust against background .
the space - fixed axial configuration is least affected by such backgrounds . * for an axial xenon polar detector , the decrease of sensitivity due to the effect of the energy threshold is much smaller than the decrease of the total event rate . a @xmath1 or @xmath152 kg@xmath2year exposure can reach a @xmath72 directional signal on a space- or earth - fixed basis with 10 kev energy threshold , for a @xmath153 wimp - nucleon cross - section and a 30 gev wimp mass . the author would like to thank adam para , jonghee yoo , yeongduk kim and stephen olsen for useful discussions .
this work was supported by the institute for basic science ( korea ) under project code ibs - r016-d1 .
99 k. a. olive _ et al . _ [ particle data group collaboration ] , `` review of particle physics , '' chin .
c * 38 * , 090001 ( 2014 ) .
d. s. akerib _ et al .
_ [ lux collaboration ] , `` first results from the lux dark matter experiment at the sanford underground research facility , '' phys
. rev .
lett . * 112 * , no . 9 , 091303 ( 2014 ) [ arxiv:1310.8214 [ astro-ph.co ] ] .
j. h. davis , `` fitting the annual modulation in dama with neutrons from muons and neutrinos , '' phys . rev . lett . * 113 * , 081302 ( 2014 ) [ arxiv:1407.1052 [ hep - ph ] ] .
d. n. spergel , `` the motion of the earth and the detection of wimps , '' phys .
d * 37 * , 1353 ( 1988 ) .
a. m. green and b. morgan , `` optimizing wimp directional detectors , '' astropart .
* 27 * , 142 ( 2007 ) [ astro - ph/0609115 ] .
j. d. vergados and a. faessler , `` direct wimp detection in directional experiments , '' phys .
d * 75 * , 055007 ( 2007 ) [ hep - ph/0611230 ] .
a. bandyopadhyay and d. majumdar , `` on diurnal and annual variations of directional detection rates of dark matter , '' astrophys .
j. * 746 * , 107 ( 2012 ) [ arxiv:1006.3231 [ hep - ph ] ] .
g. j. alner , h. araujo , a. bewick , s. burgos , m. j. carson , j. c. davies , e. daw and j. dawson _
et al . _ , `` the drift - ii dark matter detector : design and commissioning , '' nucl .
instrum .
a * 555 * , 173 ( 2005 ) .
j. b. r. battat , c. deaconu , g. druitt , r. eggleston , p. fisher , p. giampa , v. gregoric and s. henderson _
_ , `` the dark matter time projection chamber 4shooter directional dark matter detector : calibration in a surface laboratory , '' nucl
. instrum .
a * 755 * , 6 ( 2014 ) .
k. miuchi , h. nishimura , k. hattori , n. higashi , c. ida , s. iwaki , s. kabuki and h. kubo _ et al .
_ , `` first underground results with newage-0.3a direction - sensitive dark matter detector , '' phys . lett .
b * 686 * , 11 ( 2010 ) [ arxiv:1002.1794 [ astro-ph.co ] ] .
d. santos , g. bosson , j. l. bouly , o. bourrion , c. fourel , o. guillaudin , j. lamblin and f. mayet _ et al .
_ , `` mimac : micro - tpc matrix of chambers for dark matter directional detection , '' j. phys . conf .
* 469 * , 012002 ( 2013 ) [ arxiv:1311.0616 [ physics.ins-det ] ] .
s. e. vahsen , m. t. hedges , i. jaegle , s. j. ross , i. s. seong , t. n. thorpe , j. yamaoka and j. a. kadyk _ et al .
_ , `` 3-d tracking of nuclear recoils in a miniature time projection chamber , '' arxiv:1407.7013 [ physics.ins-det ] .
t. naka , m. natsume , k. niwa , k. hoshino , m. nakamura , t. nakano and o. sato , `` development of emulsion track expansion techniques for optical - microscopy - observation of low - velocity ion tracks with ranges beyond optical resolution limit , '' nucl .
instrum .
a * 581 * , 761 ( 2007 ) .
b. morgan , a. m. green and n. j. c. spooner , `` directional statistics for wimp direct detection , '' phys . rev .
d * 71 * , 103507 ( 2005 ) [ astro - ph/0408047 ] .
b. morgan and a. m. green , `` directional statistics for wimp direct detection .
2d read - out , '' phys .
d * 72 * , 123501 ( 2005 ) [ astro - ph/0508134 ] .
h. sekiya , m. minowa , y. shimizu , y. inoue and w. suganuma , `` measurements of anisotropic scintillation efficiency for carbon recoils in a stilbene crystal for dark matter detection , '' phys .
b * 571 * , 132 ( 2003 ) [ astro - ph/0307384 ] .
h. sekiya , m. mionwa , y. shimizu , w. suganuma and y. inoue , `` dark matter search with direction sensitive scintillators , '' astro - ph/0411215 .
d. r. nygren , `` columnar recombination : a tool for nuclear recoil directional sensitivity in a xenon - based direct detection wimp search , '' j. phys .
* 460 * , 012006 ( 2013 ) .
j. billard , f. mayet , j. f. macias - perez and d. santos , `` directional detection as a strategy to discover galactic dark matter , '' phys .
b * 691 * , 156 ( 2010 ) [ arxiv:0911.4086 [ astro-ph.co ] ] .
c. j. copi , l. m. krauss , d. simmons - duffin and s. r. stroiney , `` assessing alternatives for directional detection of a wimp halo , '' phys .
d * 75 * , 023514 ( 2007 ) [ astro - ph/0508649 ] .
j. billard , f. mayet and d. santos , `` assessing the discovery potential of directional detection of dark matter , '' phys .
d * 85 * , 035006 ( 2012 ) [ arxiv:1110.6079 [ astro-ph.co ] ] .
j. billard , `` comparing readout strategies to directly detect dark matter , '' phys .
d * 91 * , no . 2 , 023513 ( 2015 ) [ arxiv:1411.5946 [ physics.ins-det ] ] .
p. gondolo , `` recoil momentum spectrum in directional dark matter detectors , '' phys
. rev .
d * 66 * , 103513 ( 2002 ) [ hep - ph/0209110 ] .
g. cowan , k. cranmer , e. gross and o. vitells , `` asymptotic formulae for likelihood - based tests of new physics , '' eur .
j. c * 71 * , 1554 ( 2011 ) [ arxiv:1007.1727 [ physics.data-an ] ] .
k. s. cranmer , `` kernel estimation in high - energy physics , '' comput .
commun . * 136 * , 198 ( 2001 ) [ hep - ex/0011057 ] .
j. d. lewin and p. f. smith , `` review of mathematics , numerical factors , and corrections for dark matter experiments based on elastic nuclear recoil , '' astropart .
phys . * 6 * , 87 ( 1996 ) .
m. yamashita , t. doke , k. kawasaki , j. kikuchi , s. suzuki , `` scintillation response of liquid xe surrounded by ptfe reflector for gamma rays , '' nucl .
instrum .
a * 535 * , 692 ( 2004 ) .
j. renner _
[ next collaboration ] , `` ionization and scintillation of nuclear recoils in gaseous xenon , '' arxiv:1409.2853 [ physics.ins-det ] .
s. cebrian , t. dafni , e. ferrer - ribas , i. giomataris , d. gonzalez - diaz , h. gomez , d. c. herrera and f. j. iguaz _ et al .
_ , `` micromegas - tpc operation at high pressure in xenon - trimethylamine mixtures , '' jinst * 8 * , p01012 ( 2013 ) [ arxiv:1210.3287 [ physics.ins-det ] ] .
c. silva , j. pinto da cunha , a. pereira , v. chepel , m. i. lopes , v. solovov and f. neves , `` reflectance of polytetrafluoroethylene ( ptfe ) for xenon scintillation light , '' j. appl .
* 107 * , 064902 ( 2010 ) [ arxiv:0910.1056 [ physics.ins-det ] ] . v. m. gehman , a. goldschmidt , d. nygren , c. a. b. oliveira and j. renner , `` a plan for directional dark matter sensitivity in high - pressure xenon detectors through the addition of wavelength shifting gaseous molecules , '' jinst * 8 * , c10001 ( 2013 ) .
d. gonzalez - diaz , f. aznar , j. castel , s. cebrian , t. dafni , j. a. garcia , h. gomez and d. c. herrera _ et al .
_ , `` lessons from the operation of the `` penning - fluorescent '' tpc and prospects , '' arxiv:1504.03640 [ physics.ins-det ] .
e. ruiz - choliz , d. gonzalez - diaz , a. diago , j. castel , t. dafni , d. c. herrera , f. j. iguaz and i. g. irastorza _ et al .
_ , `` modelling the behaviour of microbulk micromegas in xenon / trimethylamine gas , '' arxiv:1506.05077 [ physics.ins-det ] .
y. nakajima , a. goldschmidt , h. s. matis , d. nygren , c. oliveira and j. renner , `` measurement of scintillation and ionization yield with high - pressure gaseous mixtures of xe and tma for improved neutrinoless double beta decay and dark matter searches , '' arxiv:1505.03585 [ physics.ins-det ] .
d. c. herrera , s. cebrin , t. dafni , j. a. garca , j. g. garza , a. goldschmidt , d. gonzalez - diaz and f. j. iguaz _ et al .
_ , `` study of columnar recombination in xe+trimethylamine mixtures using a micromegas - tpc , '' pos tipp * 2014 * , 057 ( 2014 ) .
u. fano , `` ionization yield of radiations .
the fluctuations of the number of ions , '' phys .
rev . * 72 * , 26 ( 1947 ) .
v. alvarez _ et al . _
[ next collaboration ] , `` near - intrinsic energy resolution for 30 to 662 kev gamma rays in a high pressure xenon electroluminescent tpc , '' nucl .
instrum .
a * 708 * , 101 ( 2013 ) [ arxiv:1211.4474 [ physics.ins-det ] ] .
|
we report a systematic study on the directional sensitivity of a direct dark matter detector that detects the polar angle of a recoiling nucleus . a weakly interacting massive particle ( wimp)-mass independent method is used to obtain the sensitivity of a general detector in an isothermal galactic dark matter halo . by using two - dimensional distributions of energy and polar angle ,
a detector without head - tail information with 6.3 times the statistics is found to achieve the same performance level as a full three - dimensional tracking dark matter detector .
optimum operation orientations are obtained for various experimental configurations , with detectors that are space- or earth - fixed , have head - tail capability or not , and use energy information or not .
earth - fixed detectors are found to have best sensitivity when the polar axis is oriented at a 45 degree angle from the earth s pole . with background contamination that mimics the wimp signal s energy distribution ,
the performance is found to decrease at a rate less than the decrease of signal purity .
the wimp - mass dependence of the performance of a detector with various energy thresholds that uses gaseous xenon as target material is reported .
we find that with a @xmath0 spin - independent wimp - nucleon cross - section and a 30 gev wimp , a @xmath1 kg@xmath2year s exposure with a polar detector of 10 kev threshold can make a three sigma discovery of directional wimps in the isothermal galactic dark matter halo . for a columnar recombination detector ,
experimental considerations are discussed .
|
insb(001 ) substrates ( @xmath10-type , carrier concentration of 2@xmath610@xmath11 @xmath12 ) were cleaned by repeated cycles of sputtering and annealing up to 680 k until the @xmath15(8@xmath62 ) low - energy electron diffraction ( leed ) pattern was observed .
then , a nominal 3 monolayers ( ml ) of bi were evaporated at room temperature : 1 ml is defined as the atom density of bulk - truncated insb(001 ) and the coverage was estimated by using a quartz microbalance . a subsequent flash by direct current heating up to 680 k for @xmath1310 seconds produced the @xmath14(1@xmath63 ) surface , as indicated by the sharp and low - background leed pattern shown in fig .
1 ( a ) of the main text . as shown in fig .
s1 ( a ) , the excess bi was removed after the flash . from the height of bi 5@xmath42 with linear - background subtraction ,
the coverage of bi on the ( 1@xmath63 ) surface was @xmath131.5 ml . sometimes , we could find faint fractional spots showing the @xmath15(2@xmath66 ) periodicity on the flashed surface , which suggested a small fraction of the coexisting area with the @xmath15(2@xmath66 ) periodicity , as shown in fig .
s1 ( b ) . the previous work @xcite reported a bi / insb(001)-@xmath15(2@xmath66 ) surface prepared by 2 ml of bi evaporation followed by annealing at @xmath13 470 k for 2 hours .
this surface was reported to be metallic based on the angle - integrated photoelectron spectrum taken at room temperature ( rt ) . for comparison
, we prepared the @xmath15(2@xmath66 ) surface according to ref .
the leed pattern showed the @xmath15(2@xmath66 ) periodicity ( see fig .
s1 ( c ) ) . as shown in fig .
s2 ( a ) , the fermi contour measured by arpes did not show any one - dimensional ss .
instead , there was a two - dimensional metallic state .
the angle - integrated spectra in fig . s2 ( b ) qualitatively agrees with that in the previous work .
figure s4 shows the angle - integrated photoelectron spectra taken with four different photon energies .
we determined the energy position of vbm to be the intersection between the linear fits of the background and the linear portion of the valence band leading edge @xcite .
based on the so - called linear method , the valence band maximum was estimated to be [email protected] ev .
we checked the dependence of the power index to the energy window as shown in fig .
fitting was performed based on eq .
( 1 ) in the main text . below 0.1 ev ,
the power index was almost the same and the fitting curve agreed well with the spectrum . with higher binding energies , the power index tended to become smaller and eq .
( 1 ) failed to reproduce the experimental data .
this must be because the tll framework is only valid in the low - energy region .
in addition , this deviation from the theoretical curve would be because the bottom of the ss was @xmath13 0.25 ev : around the bottom of the surface - state band , the dispersion is no longer linear .
figure s6 shows the universal scaling plot of the angle - integrated photoelectron spectra with different scaling factors @xmath4 .
fitting of the spectra taken at various temperatures to the universal curve was optimized at @xmath4 = 0.75 .
( a ) core - level spectrum for the as - deposited and
flashed bi / insb(001 ) surfaces taken at @xmath7 = 40.8 ev ( he ii ) .
( b ) leed pattern of a flashed bi / insb(001 ) surface with coexisting @xmath15(2@xmath66 ) spots ( indicated by red markers ) .
the pattern is distorted because of the flat mcp screen .
( c ) leed pattern of the bi / insb(001)-@xmath15(2@xmath66 ) surface .
, width=302 ]
|
a 1d metallic surface state was created on an anisotropic insb(001 ) surface covered with bi .
angle - resolved photoelectron spectroscopy ( arpes ) showed a 1d fermi contour with almost no 2d distortion .
close to the fermi level ( @xmath0 ) , the angle - integrated photoelectron spectra showed power - law scaling with the binding energy and temperature .
the arpes plot above @xmath0 obtained thanks to thermally broadened fermi edge at room temperature showed a 1d state with continuous metallic dispersion across @xmath0 and power - law intensity suppression around @xmath0 .
these results strongly suggest a tomonaga - luttinger liquid on the bi / insb(001 ) surface . the fermi liquid theory of ordinal three - dimensional ( 3d ) metals breaks down in one - dimensional ( 1d ) systems to produce various exotic quantum phases .
tomonaga - luttinger liquid ( tll ) @xcite is an exactly solvable model of a gapless 1d quantum system , that is characterized by power - law scaling and spin - charge separation for low - energy excitation spectra @xcite .
so far , various 1d systems have been studied as tll candidates . only a few have shown metallic states with power - law spectral features , e.g. , carbon nanotubes @xcite and lithium purple bronze @xcite .
the other characteristic of spin - charge separation has been reported in some 3d materials with 1d electronic structures @xcite .
the surface of semiconductors is known to show various self - assembled 1d atomic structures that are regarded as suitable systems for studying 1d metallic states @xcite , such as in / si(111 ) @xcite , au / si(557 ) @xcite , and pt / ge(001 ) @xcite . on such surface systems
, @xmath1-@xmath2 electron / hole doping can be performed by the deposition of additional atoms , and the local atomic structure can be observed / controlled by scanning probe techniques .
these manipulations on the surface 1d states can provide further insight into 1d physics .
hence , surface tlls have attracted much attention in the past two decades . however , despite the various 1d surfaces , most of them do not behave as tlls at low temperature .
possible reasons for the non - tll behavior have been discussed for each specific system , such as a metal - insulator transition at low temperatures @xcite and interaction with other two - dimensional ( 2d ) surface bands @xcite .
so far , the only candidate for a surface tll is an au nanowire assembled on ge(001 ) ( au / ge(001 ) ) , which shows power - law scaling for low - energy spectral features according to scanning tunneling spectroscopy ( sts ) and angle - resolved photoelectron spectroscopy ( arpes ) @xcite .
however , a recent work presented doubt with regard to the tll picture and proposed an alternative interpretation of the sts spectra : two different and gapped states lying just above / below the fermi level ( @xmath0 ) @xcite .
this picture can also explain the sts spectral shape around @xmath0 . in order to distinguish these two possibilities without ambiguity ,
the 1d band dispersion needs to be traced across @xmath0 continuously with @xmath3 resolution in reciprocal space .
however , to the best of our knowledge , no such experimental data have been reported about surface tlls so far . while some arpes results have been given for au / ge(001 ) @xcite ,
the dispersion above @xmath0 is not accessible with these data . in this letter
, we report on the case of a new surface tll candidate discovered with arpes on a bi - induced anisotropic structure on insb(001 ) ( bi / insb(001 ) ) .
the bi / insb(001 ) surface state showed a 1d fermi contour ( fc ) with almost no undulation and stayed metallic down to 35 k. the 1d feature is the unique contour on @xmath0 of the sample , which showed a parabolic dispersion with its bottom at the center of the surface brillouin zone ( sbz ) .
the photoelectron spectra in the proximity of @xmath0 obey power - law scaling as a function of binding energy and sample temperature with the same power index @xmath4 = 0.7 @xmath50.1 .
the arpes plot divided by the fermi distribution function at room temperature ( rt ) indicated the 1d state with continuous metallic dispersion across @xmath0 and power - law intensity suppression around @xmath0 .
these results strongly suggest that the 1d surface structure of bi / insb(001 ) hosts a surface tll .
( color online ) .
( a ) low - energy electron diffraction patterns of bi / insb(001)-(1@xmath63 ) taken at room temperature ( rt ) .
( b ) surface and bulk brillouin zones of bi / insb(001 ) .
( c - e ) arpes results at rt measured with @xmath7 = 21 ev .
( c ) constant energy contour around the fermi level ( @xmath0 ) .
the dashed lines indicate boundaries of the surface brillouin zone ( sbz ) . @xmath8 and @xmath9 are defined parallel and perpendicular , respectively , to [ 110 ] .
arpes intensities are integrated over 20 mev windows centered at @xmath0 .
( d ) momentum distribution curves in second sbz . ( e )
band dispersion around @xmath0 along @xmath9 .
, width=302 ] ( color online ) ( a , b ) band dispersions along @xmath9 at 8 k with @xmath7 = 15 ev .
the dotted ( solid ) curves represent the upper ( lower ) edges of projected bulk valence ( conduction ) bands .
, width=302 ] the bi / insb(001 ) surface was prepared by evaporating three monolayers of bi on the clean surface of insb(001 ) substrates ( @xmath10-type , carrier concentration of 2@xmath610@xmath11 @xmath12 ) and a subsequent flash by direct current heating up to 680 k for @xmath1310 seconds .
after the preparation , the @xmath14(1@xmath63 ) surface periodicity was observed by low - energy electron diffraction ( fig .
1(a ) ) .
the detailed procedure on sample preparation and characterization , especially the comparison with the @xmath15(2@xmath66 ) surface reported in a previous work @xcite , is shown in the supplemental material ( sm ) @xcite .
figs .
1(c)(e ) indicate the surface electronic structure of bi / insb(001 ) measured with arpes .
arpes measurements were performed with a he lamp and synchrotron radiations at the cassiope beamline of synchrotron soleil and bl7u of uvsor - iii .
the photon energies used ranged from 12 to 40.8 ev .
the photoelectron kinetic energy at @xmath0 and the overall energy resolution of each arpes setup were carefully calibrated by the fermi edge of the photoelectron spectra from freshly evaporated gold on the sample after the measurement ( see fig .
3(a ) ) .
figure 1 ( c ) shows the fermi contour ( fc ) around @xmath0 as measured with linearly polarized photons . @xmath8 and
@xmath9 are defined to be parallel to [ 110 ] and [ @xmath1610 ] , respectively .
the linear shape of the fc along @xmath9 clearly indicated a 1d metallic state on the surface .
such a 1d fc is consistent with the anisotropic ( 1@xmath63 ) surface periodicity indicated by leed .
the dispersion of the 1d state did not change with the photon energy ( shown in sm @xcite ) , which indicates no dispersion along @xmath17 as would be expected for a surface state .
figure 1 ( c ) also shows that there were no other metallic states on the surface .
therefore , this fc provides a unique and highly anisotropic conduction path on the surface . as shown in fig .
1(d ) , the arpes momentum distribution curves ( mdcs ) indicated that the ss was exactly 1d without any 2d distortion .
mdcs were captured in the second sbz because the ss dispersion became clearer there owing to the photoexitation matrix element effect .
the lack of 2d distortion can also be observed in the @xmath18-@xmath9 plot in fig .
1(e ) .
the dotted curve is the upper edge of the projected bulk valence bands as calculated according to ref .
@xcite .
the bulk valence band maximum position was determined with the linear method ( see sm for details @xcite ) .
note that the 1d fc appeared with a single domain , in contrast to the double - domain case of au / ge(001 ) @xcite , where the direction of the conduction path is still under debate because of the two equivalent domains rotated 90@xmath19 with each other @xcite .
this is because of the reduced symmetry of the zinc - blend crystal structure compared to group - iv semiconductors .
along the [ 001 ] direction , a iii - v semiconductor lattice is stacked alternately with group iii and v atoms ( indium and antimony for insb ) .
hence , the equivalent plane appears at every two atomic steps on the ( 001 ) surface of such crystal . in the group iv semiconductor crystal , every atomic plane along [ 001 ] is equivalent and appears with 90@xmath19 rotation at each atomic step to form the double - domain surface structure .
the 90@xmath19 rotation at each atomic step also occurs for iii - v semiconductors ; the double atomic step results in a 180@xmath19 rotation , so the steps are actually identical thanks to the two - fold rotation symmetry .
thus , the conduction path of the bi / insb(001 ) surface was clearly demonstrated to be parallel to [ 110 ] .
figure 2 shows the dispersion of the surface band on bi / insb(001 ) .
the surface state formed a single parabolic band with its bottom at @xmath8 = 0 @xmath20 .
this band is the unique metallic feature on the surface .
this 1d metallic band crossed @xmath0 only once on each side of the sbz . at higher binding energies , the bottom of the surface - state band showed a small shift ( @xmath1350 mev ) between @xmath9 = 0.0 and 0.2 @xmath20 .
this would be due to an overlap of the bulk valence bands .
the dotted ( solid ) curves in fig .
2 are the upper ( lower ) edges of the projected bulk valence ( conduction ) bands and were calculated in the same manner as those in fig .
1(e ) .
the bottom of the ss overlapped the bulk valence bands at the center of the sbz .
note that the bulk conduction band does not cross @xmath0 , as shown by the solid curve in fig .
2(a ) .
( color online ) .
( a ) photoelectron spectrum ( red markers ) integrated along @xmath8 ( @xmath9 = [email protected] @xmath20 ) . the solid line between @xmath5 0.1 ev is the fitting curve ( see text for details ) .
( b ) temperature dependence of the photoelectron intensity at @xmath0 ( @xmath21 ) .
each spectrum was normalized at the intensity at 0.2 ev below @xmath0 in order to compensate the sample degradation .
the solid ( blue ) line is the fitting curve based on power - law scaling ( @xmath22 ) convolved with @xmath23 ( 10 mev ) .
the dashed line is @xmath24 without broadening from @xmath23 .
( c ) the same as ( b ) plotted in a log scale .
( d ) universal scaling plot of the spectra at various temperatures , using a @xmath25-renormalized energy scale ( ev/@xmath26 ) .
the thick dotted ( green ) line is from eq .
( 1 ) with the scaling factor @xmath4 = 0.75 .
, width=302 ] in order to examine the 1d nature of the surface state , we compared the angle - integrated photoelectron spectrum with that from ordinary metal ( au evaporated on the sample after the arpes measurement ) , as shown in fig .
3 ( a ) .
the angle integration was performed at @xmath9 = [email protected] @xmath20 .
figure 3 ( a ) clearly shows the suppression of the photoelectron intensity at @xmath0 compared to that from au , which suggests a deviation of the surface state from the fermi - liquid framework .
we tried to fit the spectrum to the tll spectral function considering the finite temperature , as given in some of the literature @xcite .
based on ref .
@xcite , the spectral function of the angle - integrated pes can be written as , @xmath27 where @xmath28 = @xmath29 is the temperature - normalized energy , @xmath30 is the gamma function , and @xmath31 is the fermi distribution function . when convolved with the gaussian instrumental energy resolution ( @xmath23 = 10 mev ) , this function reproduced the spectra well at 35 k ( fig .
3 ( a ) ) with @xmath4 = 0.65 @xmath5 0.05 in the low - energy region .
this function only fits the spectrum below 0.1 ev .
this was because the tll framework is only valid in the proximity of @xmath0 ( see sm for detailed analysis @xcite ) .
the power - law scaling of a tll should show an universal power index with various parameters . figs .
3 ( b ) and ( c ) show the temperature dependence of the photoelectron intensities at @xmath0 .
the intensity from 180 to 50 k decreased linearly with log @xmath25 , which indicates power - law scaling of the spectral intensities , @xmath32 .
the slight change in the slope below 50 k was due to the finite @xmath23 .
this broadening was already convolved in the fitting curve ( blue solid line ) to give a power index of @xmath33 , which is a value very close to that obtained by fitting the energy scale ( fig .
3 ( a ) ) . note that @xmath21 should be constant in the case of a fermi - dirac system .
another method to examine universal power - law scaling is to normalize the spectra taken at various temperatures to @xmath34 and plot them versus the temperature - normalized energy @xmath28 , as shown in fig .
3 ( d ) . with such a renormalization ,
all spectra with various @xmath25 become almost identical in the low - energy region , typically in @xmath35 5 . for such identification , the scaling factor @xmath4 = 0.75 @xmath5 0.05 is required ( spectra renormalized with another @xmath4 are shown in the sm @xcite ) . because such a temperature normalization method can include the intensity not only at @xmath18=@xmath0 but within the finite energy window , this renormalized plot would be the best way to estimate the power index from angle - integrated spectra taken at various temperatures .
all of these spectral features indicate that the 1d surface state on bi / insb(001 ) obeys an universal power - law scaling with @xmath4 = [email protected] , which agrees with that expected for a tll .
( color online ) .
( a ) arpes intensity plot along @xmath8 divided by the fermi distribution function convolved with the instrumental resolution taken at @xmath7 = 21 ev .
arpes data were integrated along @xmath9 within @xmath36 0.3 @xmath20 for better statistics .
( b ) arpes mdcs along @xmath8 within 20 mev energy window .
the intensities of each curve were normalized by the peak height .
solid lines are the fitting curves ( see text for details ) . ( c ) an angle - integrated edc within @xmath37 0.3 @xmath20 divided by the fermi distribution function , similar to ( a ) .
, width=302 ] to obtain further insight into the 1d ss , we looked into the detailed structure of arpes spectra .
figure 4 ( a ) shows the arpes intensity plot at rt along @xmath8 divided by the fermi distribution function convolved with the instrumental resolution . for better statistics , the arpes plot was integrated along @xmath9 within @xmath360.3 @xmath20 .
the momentum distribution curves ( mdcs ) are shown in fig .
4 ( b ) .
they are fitted well by a pair of voigt peaks ( gaussian and lorentzian widths are fixed for every peak ) and linear backgrounds ( solid lines in fig .
4 ( b ) ) . weak residuals around @xmath8 = 0 @xmath20 at the binding energies below -0.1 ev would be due to the bulk conduction bands . from the mdcs
, we obtained the peak positions at each binding energy and overlaid them on fig .
4 ( a ) .
they indicated the fermi wavevector @xmath38 to be 0.05 @xmath50.02 @xmath20 .
this estimation is consistent with the arpes data taken at low temperature shown in fig .
2 . this shows the continuous , metallic dispersion of the 1d ss across @xmath0 .
moreover , the arpes intensity plot clearly shows that the spectral intensity of the 1d ss was suppressed around @xmath0 , without breaking the continuity of the band dispersion . such suppression of the density of states ( dos ) around @xmath0 without any gap opening reflects the power - law scaling of the tll .
this result excludes alternative possibilities without ambiguity .
two different states below / above @xmath0 can not reproduce such a dispersion .
in addition , the angle - integrated spectrum showed a temperature - broadened power - law shape for the density of states around @xmath0 , as expected in theory @xcite . at rt
, it sometimes becomes difficult to determine the exact position of @xmath0 because of the broad fermi edge .
however , note that such ambiguity did not affect the dos suppression obtained by arpes : the intensity suppression appeared irrespective of slight artificial shifts in @xmath0 . in summary , a new candidate for a surface tll formed on bi / insb(001 ) has been discovered .
a surface state formed a 1d fc with almost no 2d undulation and stayed metallic down to 35 k. this 1d fc is the only metallic feature of the sample and hence provides an unique 1d conduction path on the surface .
the photoelectron spectra in the proximity of @xmath0 obeyed the power - law scaling as a function of binding energy and sample temperature with the same power index @xmath4 = 0.7 @xmath50.1 .
moreover , the arpes plot above @xmath0 obtained thanks to thermally broadened fermi edge at rt indicated a 1d state with continuous metallic dispersion across @xmath0 and power - law intensity suppression around @xmath0 .
these results strongly suggest that the 1d surface structure bi / insb(001 ) hosts a surface tll .
the new 1d ss on bi / insb(001 ) would provide a fertile playground for further studies of low - dimensional physics .
we thank v. meden for the helpful discussions .
we also acknowledge d. ragonnet and f. deschamps for their support during the experiments on the cassiope beamline at synchrotron soleil .
part of the arpes experiments were performed under the uvsor proposal no .
26 - 824 and nanotechnology platform program of the ministry of education , culture , sports , science and technology ( mext ) , japan .
this work was also supported by the jsps grant - in - aid for research activity start - up ( grant no .
26887024 ) .
99 s. tomonaga , prog .
. phys . * 5 * , 544 ( 1950 ) .
j. m. luttinger , j. math . phys . * 4 * , 1154 ( 1963 ) .
j. voit , rep .
. phys . *
57 * , 977 ( 1994 ) .
h. ishii @xmath39 @xmath40 , nature * 426 * , 540 ( 2003 ) .
f. wang @xmath39 @xmath40 , phys .
rev .
lett . * 96 * , 196403 ( 2006 ) .
l. dudy @xmath39 @xmath40 , j. phys . : condens .
matt . * 25 * , 014007 ( 2013 ) .
r. claessen @xmath39 @xmath40 , phys .
rev .
lett . * 88 * , 096402 ( 2002 ) .
b. j. kim @xmath39 @xmath40 , nat .
phys .
* 2 * , 397 ( 2006 ) .
y. jompol @xmath39 @xmath40 , science * 325 * , 597 ( 2009 ) . f. j. himpsel @xmath39 @xmath40 , j. phys . : condens
. matt . * 13 * , 11097 ( 2001 ) .
m. grioni , s. pons and e. frantzeskakis , j. phys . : condens .
matt .
* 21 * , 023201 ( 2009 ) .
j. nogami , s. park and c. f. quate , phys .
rev .
b * 36 * , 6221 ( 1987 ) .
p. segoviaa , d. purdie , m. hengsberger and y. baer , nature * 402 * , 504 ( 1999 ) .
o. gurlu , o. a. o. adam , h. j. w. zandvliet and b. poelsema , appl .
phys . lett . *
83 * , 4610 ( 2003 ) . h. w. yeom @xmath39 @xmath40 , phys .
rev .
lett .
* 82 * , 4898 ( 1999 ) . k. yaji @xmath39 @xmath40 , phys .
rev .
b * 87 * , 241413(r ) ( 2013 ) . c. blumenstein @xmath39 @xmath40 , nat . phys .
* 7 * , 776 ( 2011 ) .
s. meyer @xmath39 @xmath40 , phys .
rev .
b * 90 * , 125409 ( 2014 ) .
j. park @xmath39 @xmath40 , phys .
rev .
b * 90 * , 165410 ( 2014 ) .
k. nakatsuji , y. motomura , r. niikura and f. komori , phys .
rev .
b * 84 * , 115411 ( 2011 ) .
p. laukkanen @xmath39 @xmath40 , phys .
rev .
b * 81 * , 035310 ( 2010 ) .
see supplemental material [ url ] , which includes refs .
@xcite .
s. a. chambers , t. droubay , t. c. kaspar and m. gutowski , j. vac .
sci .
technol .
b * 224 * , 2205 ( 2004 ) .
d. j. chadi , phys . rev .
b * 16 * , 790 ( 1977 ) .
k. yaji @xmath39 @xmath40 , @xmath41 k. nakatsuji and f. komori , nat . phys .
* 8 * , 174 ( 2012 ) .
k. schnhammer and v. meden , j. electron .
spectrosc .
relat . phenom . *
62 * , 225 ( 1993 ) d. orgad , philos . mag .
b * 81 * , 377 ( 2001 ) .
|
@xmath0 centauri ( ngc 5139 ) is , to the best of our knowledge , the most massive globular cluster in the milky way , with a mass around @xmath3 @xcite .
furthermore , it shows peculiar chemical and dynamical features , such as the wide spread in metallicity distribution of member stars @xcite and the difference in kinematics of metal - rich and metal - poor populations @xcite .
some member stars show an enhancement of s - process elements produced by agb stars ( e.g. , * ? ? ?
* ) , which means that the potential well must have been deep enough to trap the gas ejected from agb stars @xcite .
a merger model @xcite can explain the kinematics and metallicity spread , but has difficulty in accounting for the presence of s - process elements .
a tidal stripping scenario @xcite has been proposed to explain all observed features simultaneously . in this scenario ,
the cluster was born as a dwarf elliptical galaxy , which was much more massive than the current @xmath0 centauri .
as it sinks to the galactic center through dynamical friction , it loses most of its mass through the tidal stripping by the gravitational potential of the milky way .
however , if the central density is sufficiently high , the core of the dwarf can still survive as a bound system and become a relatively large globular cluster , something like @xmath0 centauri . in this scenario , a relatively long star formation history
is naturally explained because the potential well of the progenitor dwarf elliptical was initially much deeper than that of the present - day @xmath0 centauri .
in addition , since the progenitor dwarf could be formed through hierarchical merging events , signatures of past merging events can also be explained .
furthermore , recently , @xcite reported that some of massive globular clusters in ngc 5128 may also be such tidally stripped dwarf ellipticals . from the viewpoint of dynamics , whether or not the stripping scenario works depends on the following two questions .
the first one is whether the dynamical friction can bring the progenitor dwarf to the current orbit of @xmath0 centauri .
the second one is whether the tidal stripping can actually produce the observed surface brightness of @xmath0 centauri .
the first question has been addressed by recent work @xcite .
however , the second question has not been studied so far , despite its significance to the validity of the stripping scenario . in this paper
, we have investigated whether a dwarf galaxy with the orbit similar to that of the present - day @xmath0 centauri will evolve to have the spatial structure which agrees with that of @xmath0 centauri using @xmath1-body simulations . in 2 , numerical models and method are described .
results are shown in 3 .
a discussion is given in
we consider the dynamical evolution of a dwarf elliptical in a fixed potential . in this paper , we neglect the effects of dynamical friction .
this neglect should not affect results as far as a progenitor dwarf along the present - day orbit of @xmath0 centauri is concerned , since our main goal is to see whether or not the tidal field of the milky way in the present - day orbit of @xmath0 centauri can account for its density structure .
in reality , dynamical friction must have played a certain role in bringing the progenitor dwarf to its current orbit , but that part of evolution is not in the scope of the present paper .
we use the data compiled by @xcite for the surface - brightness profile of @xmath0 centauri .
this profile is well - fitted ( except in the outermost region ) by a king model with a non - dimensional central potential @xmath4 and a core radius @xmath5 .
we choose the total mass of this king model to be identical to the mass suggested from observations , @xmath6 .
then , we set up the model of the progenitor dwarf to have the same core radius and central velocity dispersion , but with a much more extended halo .
we choose a king model with @xmath7 as a model for the dwarf .
since some dwarf elliptical galaxies can also be fitted by a king model , we believe this choice is justified . @xcite have found that a king - like dwarf elliptical can not evolve to @xmath0 centauri on the basis of numerical simulations of king- and hernquist - like dwarf galaxies embedded in a live milky way potential .
this point will be discussed in 4.1 .
the dwarf elliptical is represented by @xmath8 particles of equal mass .
the initial total mass is @xmath9 .
we do not include a dark matter halo of the dwarf , which would be more extended than the distribution of visible stars .
since the extended halo would be stripped out anyway , we believe the presence of a dark matter halo does not change results significantly . in most of our simulations ,
the final cluster still had more than 50,000 particles , with the result that the relaxation time is much longer than the duration of the simulation .
we use two different models for the gravitational potential of the milky way , a singular isothermal sphere ( run std ) and a bulge - disk - halo model ( run mwd ) .
the circular velocity of the std potential is @xmath10 .
the mwd potential is almost the same as the model used in @xcite .
this model consists of three components : a miyamoto - nagai @xcite disk , a hernquist @xcite bulge , and a logarithmic halo potential .
the only difference between the model of @xcite and ours is the core radius of the logarithmic potential .
we set the core radius to be @xmath11 , while @xcite set it to be @xmath12 , so that the circular velocity at the solar circle radius , i.e. , @xmath13 , is @xmath14 .
the potentials and parameters used in simulations are listed in table [ tab : model ] .
the circular speed curves of these models are plotted in figure [ fig : model ] .
we use the following kinematical data on @xmath0 centauri : the distance from the sun is @xmath15 , the proper motions are @xmath16 , and the radial velocity is @xmath17 @xcite .
for the most probable values of the proper motions and the radial velocity in the isothermal model ( run std ) , the pericenter and apocenter distances are @xmath18 and @xmath19 , respectively .
we made two other runs , in which we expand the initial dwarf by a factor of 1.5 ( run r15 ) and 2 ( run r20 ) , to see the effect of the change in the tidal force at the pericenter .
for the singular isothermal model , the only characteristic scale in the experiment is the core radius of the cluster .
a linear expansion of the cluster by a factor of 1.5 or 2 can therefore equally well correspond to a reduction in the pericenter distance of the unmodified cluster by a factor of @xmath20 ( run r15 ) or @xmath21 ( run r20 ) , which corresponds to @xmath22 and @xmath23 , respectively .
we use a hierarchical tree algorithm @xcite on the grape-6 hardware @xcite with an opening angle being 0.5 .
we employ a plummer softening of @xmath24 , which is roughly one - tenth of the observed core radius of @xmath0 centauri , @xmath25 .
the equations of motion are integrated using a leap - frog method with a constant time - step of @xmath26 .
this time - step is comparable to the time required for a particle with the maximum circular velocity around @xmath0 centauri to cover the softening length .
the total energy was conserved to better than @xmath27 per cent in all simulations .
figure [ fig : snap ] shows snapshot images from run std , along with the orbit of the cluster center for the periods of 0.1 gyr before and after the time for the snapshot ( whenever the orbit is available ) .
the cluster center is defined as the position of the particle with the minimum potential energy for the @xmath1-body particles .
we include all particles to calculate the potential . removing unbound particles
did not affect the result , even after more than 90 per cent of the particles were stripped .
we can see that the stripped stars remain close to the cluster orbit , and form numerous ripple structures .
figure [ fig : mass ] shows the time evolution of the cluster mass . here
, we define the cluster mass simply as a mass within @xmath28 , where @xmath29 is the tidal radius at the apocentric position expressed as @xmath30 here @xmath31 is the distance from the galactic center to the cluster , @xmath32 is the galaxy mass within @xmath31 , and @xmath33 is the cluster mass .
we assume that the stars are stripped from the cluster if their distances from the cluster center are larger than two apocentric tidal radii ( e.g. , * ? ? ?
for the three - component model , we have determined the cluster mass in the same way .
the cluster mass exhibits periodic sudden drops , which correspond to the pericentric passage .
thus , almost all the mass loss occurred at the pericentric passage .
since the cluster is expanded in runs r15 and r20 , the mass loss is larger for these runs than that for run std .
there is very little difference between the result of run std and that of mwd , suggesting that the disk does not contribute significantly to the tidal force .
this result is of course not surprising , because the disk gravity is small at a pericenter distance of @xmath34 .
figure [ fig : sden ] shows the final surface - density profiles for all runs .
we assume a constant mass - to - light ratio , since the estimated relaxation time of @xmath0 centauri is too long for the mass segregation to be visible ( e.g. , * ? ? ?
clearly , the result of run std does not agree with observations . replacing the galactic potential with a more realistic one ( run mwd )
does not improve the situation .
on the other hand , changing the pericenter distance has a drastic effect , and the agreement of run r20 with observations is very good .
the result of r15 falls between std and r20 .
we conclude that a tidal stripping scenario involving a king - like progenitor without a dark matter halo can reproduce the observed luminosity profile quite well , provided that the pericentric distance of @xmath0 centauri is around @xmath35 .
our conclusion that a king - like dwarf can evolve to @xmath0 centauri apparently contradicts with the result by @xcite that final clusters started from king - like dwarf ellipticals are still more massive than @xmath36 .
this difference is mainly due to the high velocity dispersion of the initial dwarf models in their simulations .
the central velocity dispersion in their model k5 reaches @xmath37 , while we set it to be @xmath38 .
this means that the cluster mass inside the tidal radius in their model k5 is @xmath39 times larger than the mass in our model , even if the pericenter distance is assumed to be equal . in terms of the fraction of total mass lost ,
their results are actually quite consistent with ours .
we have shown that the observed surface - brightness profile of @xmath0 centauri is nicely reproduced by the tidal stripping scenario if the pericenter distance of the cluster is roughly @xmath35 . here
, we discuss whether such a small pericenter distance is compatible with observations .
we have adopted the following error estimates for observational data .
for the errors in the proper motion and the radial velocity , we have used the data by @xcite : @xmath40 . here , @xmath41 is the relative velocity of @xmath0 centauri to the local standard of rest .
the positive directions of the @xmath42 , @xmath43 , and @xmath44 components are , respectively , outward from the galactic center , toward the galactic rotation , and toward the galactic north pole . for the galactic parameters
, @xcite estimated the ratio of the circular speed to the solar circle radius , @xmath45 , on the basis of the proper motion study of sagittarius a@xmath46 by using vlba , and concluded that the circular speed is @xmath47 , assuming that the solar circle radius is @xmath48 .
for the error in the circular speed at the solar circle , we have adopted the value of @xmath49 .
since the solar circle radius would have an error comparable to that in the circular speed , we have used @xmath50 for the solar circle radius .
we calculated the variations of the pericentric distance due to errors in the orbital velocity of @xmath0 centauri , the solar circle radius , and the circular speed at the solar circle .
the results are summarized in table [ tab : err ] .
the error in the circular speed has a strong effect on the pericenter distance , since the orbit of @xmath0 centauri is highly eccentric .
a 10 per cent variation in the circular speed can lead to 50 per cent change in the pericenter distance .
the error in the @xmath43 component also has such large magnification .
thus , our result that the pericenter distance must be around @xmath35 is compatible with observations .
we set the total mass of @xmath0 centauri to @xmath6 , following the estimate by @xcite .
however , since this mass estimate is based on a multi - mass king model , the total mass is probably overestimated ( see * ? ? ?
* ) . indeed , in our best - fit king model , the central velocity dispersion reaches @xmath38 , while the observed value is around @xmath51 @xcite . for the observed value of the central velocity dispersion ,
the total mass is around @xmath52 .
this change of the cluster mass reduces the tidal radius by about 15 per cent ( @xmath53 ) . in other words , to yield the observed tidal radius with a reduced cluster mass , the periastron distance in the model would have to be increased by 30 per cent [ from equation ( [ eq : rt ] ) , @xmath54 . therefore ,
if we use this revised estimate of the mass of @xmath0 centauri , the most likely value of the pericenter distance is around @xmath2 . in section 3
, we have found that the quantitative result remains unchanged when we change the galactic potential model from the singular isothermal sphere to the more realistic three - component model .
however , since we integrated our model only for 0.88 gyr , we could have missed some important contributions of the non - sphericity of the potential .
we integrated the orbit of a point - mass cluster in the three - component potential for @xmath55 . here
, such a long integration time is required since chaotic diffusion time could be much longer than a hubble time .
we have found that the minimum pericentric radius is @xmath56 , although the orbit is chaotic .
consequently , the effect of chaotic diffusion is negligible .
we have found that the structure of @xmath0 centauri is nicely explained by the tidal stripping scenario , for the orbital parameters within the observational errors .
the pericenter distance is the most important parameter that determines the structure of the simulated @xmath0 centauri . in other words
, we can put a fairly tight constraint on the pericenter distance of @xmath0 centauri from its internal structure .
this constraint can be converted to constraints on the galactic potential , such as the circular speed , if we have high - accuracy data for the proper and radial motions of @xmath0 centauri .
we can probably apply a similar technique to other tidally - limited globular clusters , though in this case the modeling is somewhat more complex since we can not ignore thermal relaxation .
clusters in highly eccentric orbits are most useful , since a small change in the galactic parameters results in a large change in the pericenter distance .
thus , the dynamical simulation of the evolution of globular clusters in a galactic tidal field , combined with high - accuracy proper motion data which will be available via next - generation astrometry projects , will provide us with a unique tool to probe the structure of the milky way .
we thank the anonymous referee for useful comments that helped us to improve the clarity of the paper .
we are grateful to prof .
s. hozumi for his critical reading of the manuscript .
all simulations were run at the university of tokyo .
some part of data analysis was made on workstations at astronomical data analysis center , national astronomical observatory of japan ( adac / naoj ) .
zinnecker , h. , keable , c. j. , dunlop , j. s. , cannon , r. d. , & griffiths , w. k. 1988 , in iau symp .
126 , globular cluster systems in galaxies , ed .
j. e. grindlay & a. g. d. philip ( dordrecht : kluwer ) , 603
|
we have investigated whether or not a tidal stripping scenario can reproduce the observed surface - brightness profile of @xmath0 centauri using @xmath1-body simulations .
assuming that the progenitor of @xmath0 centauri is a dwarf elliptical galaxy , we model it with a king model with a core radius being the same as that of @xmath0 centauri .
a dark matter halo of the dwarf is not taken into account .
we consider two different models of the milky way potential : a singular isothermal sphere and a three - component model .
the progenitor dwarf is expressed as an @xmath1-body system , which orbits in the fixed galactic potential .
the dwarf lost more than 90 per cent of its mass during the first few pericenter passages .
thereafter , the mass remains practically constant .
the final surface - density profile is in good agreement with the observational data on @xmath0 centauri , if the pericenter distance of the orbit of the progenitor dwarf is around @xmath2 .
this value is within the error bar of the current proper motion data on @xmath0 centauri and galactic parameters .
although our simulation is limited to a king - like progenitor dwarf without a dark matter halo , it strongly suggests that the current density profile of @xmath0 centauri is nicely reproduced by a tidal stripping scenario , in other words , that @xmath0 centauri can plausibly be identified with a stripped dwarf elliptical .
|
modeling and simulation of multiphase fluid systems with a realistic equation of state ( e.g. peng - robinson equation of state @xcite ) has become an attractive and challenging research topic in the chemical and reservoir engineering @xcite .
it plays very important role in the pore scale modeling and simulation of subsurface fluid flow , especially shale gas reservoir that has become an increasingly important source of natural gas in the recent years .
the mathematical models of multiphase fluids are often formulated by a set of thermodynamic state variables and fluid velocity . in the traditional framework of modeling multiphase fluids ,
the thermodynamic state variables are the pressure , temperature , and chemical composition ( the so - called npt - based framework ) .
this framework has been extensively used in many applications @xcite . however , the npt - based framework has some essential limitations @xcite .
first , a realistic equation of state ( e.g. peng - robinson equation of state ) is a cubic equation with respect to the density , so the density might not be uniquely determined for the specified pressure , temperature , and molar fractions .
second , @xcite states that the specification of pressure , temperature , and mole fractions does not always determine the equilibrium state of the system uniquely .
in addition , in compositional fluid simulation , the pressure is not known a - priori and there exists no intrinsic equation for the pressure .
this leads to the complication of constructing the pressure evolution equation for application of the npt - based framework @xcite . in order to resolve the issues of the npt - based framework ,
an alternative framework @xcite has been developed , and it uses the moles , volume , and temperature ( the so - called nvt - based framework ) as the primal state variables . the nvt - based framework is initially applied to the phase - split computations of multi - component fluids at the constant moles , volume and temperature ( also called nvt flash computation ) in @xcite , and for the subsequent research progress of the nvt flash computation , we refer to @xcite .
recently , the nvt - based framework has been successfully applied in the numerical simulation of compositional two - phase flow in porous media @xcite . very recently , in @xcite , we have extended the nvt - based framework to the pore scale and developed a diffuse interface model to simulate multi - component two - phase flow with partial miscibility , but the rigorous mathematical analysis is absent to demonstrate the consistency with the thermodynamical laws . in this work
, we will propose a general mathematical model for multi - component two - phase flow with partial miscibility , which is viewed as an extension of the nvt - based framework .
moreover , the proposed model is derived with the mathematical rigor based on the thermodynamic laws and a realistic equation of state ( e.g. peng - robinson equation of state ) .
the diffuse - interface models for two - phase incompressible immiscible fluids have been developed in the literature , @xcite for instance .
the proposed model in this work can characterize the compressibility and partial miscibility . a significant contribution in @xcite
is that an additional term involving mass diffusions is introduced in the momentum equation to ensure consistency with thermodynamics for the case of non - matched densities . as shown in our derivations of section [ secmodel ] , we find that an analogical term that also results from mass diffusions is crucial to establish the thermodynamically consistent model of compressible and partially miscible multi - component two - phase flow .
this term occurs in the momentum equation through a different but natural derived form , which satisfies the thermodynamic laws .
a major distinction is that the nvt - based framework uses the helmholtz free energy rather than gibbs free energy used in the npt - based framework .
the first challenging problem caused by this transition is that the traditional techniques @xcite can no longer work well to derive the two - phase flow model .
it is well known that the entropy balance equation plays a fundamental role in the derivations of two - phase flow model . in the classical non - equilibrium thermodynamics @xcite , the variation of entropy @xmath0 exists , so the gibbs relation can be used to derive this key equation . however ,
when the helmholtz free energy is introduced in the gibbs relation , the variation of entropy @xmath0 is eliminated . as a result
, the gibbs relation can not be used to derive the entropy balance equation as in the classical thermodynamics . in this work ,
we resolve this problem by combing the first law of thermodynamics and the related physical relations , and from this , we derive consistent entropy balance equations .
different from the npt - based framework , we use a thermodynamic pressure , which is no longer an independent state variable in the nvt - based framework , and indeed , it is a function of the molar density and temperature .
consequently , it is never necessary to construct the pressure evolution equation although the velocity field is no longer divergence - free .
the proposed model is related to the dynamic van der waals theory of two - phase fluid flow for a pure substance with the thermodynamic pressure , which is developed from physical point of view @xcite .
such model is first proposed in @xcite , and it has successful applications @xcite for example .
but the proposed model has substantial differences from @xcite that we consider the fluids composed of multiple components and introduce an additional term involving mass diffusions in the momentum equation by a natural way .
moreover , we establish the relation between the pressure gradient and gradients of the chemical potentials . based on these relations , we derive a new formulation of the momentum conservation equation , which shows that the gradients of chemical potentials become the primary driving force of the fluid motion .
this formulation allows us to conveniently prove that the total ( free ) energy of the proposed model is dissipated with time . for numerical simulation ,
a main challenge of diffuse interface models is to design efficient numerical schemes that preserve the discrete ( free ) energy dissipation @xcite .
there are a lot of efforts on the developments of such schemes in the literature ; in particular , a notable progress is that efficient numerical simulation methods for simulating the model of @xcite have been proposed in @xcite .
the key difficulties encountered in constructing energy - stable numerical simulation for the proposed model are the strong nonlinearity of helmholtz free energy density and tight coupling relations between molar densities and velocity . to resolve these problems , we propose a novel convex - concave splitting of helmholtz free energy density and deal well with the coupling relations between molar densities and velocity through very careful physical observations with a mathematical rigor .
the proposed numerical scheme is proved to preserve the discrete ( free ) energy dissipation .
the key contributions of our work are listed as below : \(1 ) a general diffuse interface model with a realistic equation of state ( e.g. peng - robinson equation of state ) is proposed to describe the multi - component two - phase fluid flow based on the principles of the nvt - based framework .
\(2 ) we combine the first law of thermodynamics and the related physical relations to derive the entropy balance equations , and then we derive a transport equation of the helmholtz free energy density .
finally , using the second law of thermodynamics , we derive a set of unified equations for both interfaces and bulk phases that can describe the partial miscibility of two fluids .
\(3 ) a term involving mass diffusions is naturally included in the momentum equation to ensure consistency with thermodynamics .
\(4 ) we prove a relation between the pressure gradient and the gradients of the chemical potentials , and from this , we derive a new formulation of the momentum balance equation , which demonstrates that chemical potential gradients become the primary driving force of the fluid motion .
\(5 ) we prove that the total ( free ) energy of the proposed model is dissipated with time .
\(6 ) an energy - dissipation numerical scheme is proposed based on a convex - concave splitting of helmholtz free energy density and a careful treatment of the coupling relations between molar densities and velocity .
numerical tests are carried out to verify the effectiveness of the proposed method .
the rest of this paper is organized as follows . in section 2
, we will introduce the related thermodynamic relations . in section 3 , we derive a general model for multi - component two - phase fluid flow . in section 4 ,
we prove the total ( free ) energy dissipation law of the proposed model . in section 5
, we propose an efficient energy - stable numerical method , and in section 6 numerical tests are carried out to verify effectiveness of the proposed method . finally , concluding remarks are provided in section 7 .
we consider a fluid mixture composed of @xmath1 components under a constant temperature ( @xmath2 ) .
we denote the molar density vector by @xmath3^t$ ] , where @xmath4 is the molar density of the @xmath5th component .
furthermore , we denote the overall volume by @xmath6 , and then the moles of component @xmath5 is @xmath7 .
the nvt - based framework uses the moles of each component ( @xmath8 ) and volume ( @xmath6 ) together with the given constant temperature as the primal state variables . with the relation @xmath9
, the primal state variables can be reduced into the molar density ( @xmath10 ) . by fundamental laws of thermodynamics @xcite
, we have the gibbs relation @xcite @xmath11 where @xmath12 is the internal energy , @xmath2 is the temperature , @xmath13 is the entropy , @xmath14 is the pressure , and @xmath15 is the chemical potential of component @xmath5 .
we define the helmholtz free energy as usual : @xmath16 with the definition of the helmholtz free energy , we get @xmath17 , and thus , the gibbs relation becomes @xmath18 we first note that the entropy balance equation plays a fundamental role in the derivations of two - phase flow model . in the classical derivations @xcite ,
the gibbs relation is used to derive this key equation because it contains the variation of entropy @xmath0 .
however , we can see from that after the helmholtz free energy is introduced in the gibbs relation , the variation of entropy @xmath0 is eliminated . as a result
, the gibbs relation can not be used to derive the entropy equation as in the classical theory , and it is clearly in need of the other alternative relations . in order to resolve this problem , we will use the first law of thermodynamics and the entropy structure rather than the gibbs relation to derive the entropy balance equation .
we now define the helmholtz free energy density as @xmath19 , and then have its form @xmath20 in general , the helmholtz free energy density is a function of @xmath10 and temperature .
we introduce briefly the formulations of helmholtz free energy density @xmath21 of a homogeneous bulk fluid determined by peng - robinson equation of state , which is widely used in the oil reservoir and chemical engineering .
the proposed model can also apply the other realistic cubic equations of state , for instance , the van der waals equation of state @xcite , which is popularly used in physics .
the helmholtz free energy density @xmath21 of a bulk fluid is calculated by a thermodynamic model as @xmath22 where @xmath23 , @xmath24 and @xmath25 are formulated in the appendix .
the mole - pressure - temperature form of peng - robinson equation of state @xcite is @xmath26 which is a cubic equation of the overall molar density @xmath27 . in the mathematical sense , might have not a unique solution for the specified pressure , temperature , and molar fractions ( that is , @xmath28 are given and thus @xmath29 are known ) .
it is a well - known drawback of the npt - based framework .
however , under the nvt - based framework , can always provide a unique and explicit pressure for given molar density and temperature .
we note that for bulk fluids , the pr - eos formulation is equivalent to the pressure equation ( see the details in the appendix ) @xmath30 which results from .
however , the pressure equation is more general and it can be also used to define and calculate the pressure of an inhomogeneous fluid . for realistic fluids , the diffuse interfaces always exist between two phases . the interfacial partial miscibility is a phenomenon that the two - phase fluids behave on the interfaces . to model this feature , a local density gradient contribution is introduced into the helmholtz free energy density of inhomogeneous fluids .
the general form of helmholtz free energy density ( denoted by @xmath31 ) is then the sum of two contributions : helmholtz free energy density of bulk homogeneous fluid and a local density gradient contribution : @xmath32 where @xmath33 here , @xmath34 is the cross influence parameter .
the density gradient contribution accounts for the phase transition by the gradual density changes of each component on the interfaces .
but for a bulk phase , the molar density of each component has uniform distribution in space , and in this case , @xmath35 vanishes . for the influence parameters , it is generally considered constant in this paper ( see the appendix ) .
the chemical potential of component @xmath5 is defined as @xmath36 where @xmath37 and @xmath38 is the variational derivative .
moreover , the general pressure can be formulated as @xmath39 where @xmath40 from , the pressure is a function of @xmath10 and @xmath2 in the nvt - based framework , but it is no longer an independent state variable as it is so in the npt - based framework .
in this section , we will derive the general model for the multi - component two - phase fluid flow , in which the viscosity and density gradient contribution to free energy are under consideration .
first , we use the first law of thermodynamics and entropy splitting structure to derive an entropy equation , and then we derive a transport equation of helmholtz free energy density to further reduce the entropy equation . from the reduced entropy equation , we derive a general model of multi - component two - phase flow , which obeys the second law of thermodynamics .
the first law of thermodynamics states @xmath41 where @xmath42 is the time , @xmath43 is the kinetic energy , @xmath44 is the work done by the face force @xmath45 , and @xmath46 stands for the heat transfer from the surrounding that occurs to keep the system temperature constant .
we split the total entropy @xmath13 into a summation of two contributions .
one is the entropy of the system , denoted by @xmath47 .
the other is the entropy of the surrounding , denoted by @xmath48 , which has the relation with @xmath46 as @xmath49 taking into account the relation @xmath50 , and using and , we have @xmath51 we denote by @xmath52 the molar weight of component @xmath5 , and define the mass density of the mixture as @xmath53 in a time - dependent volume @xmath54 , we define the entropy , helmholtz free energy and kinetic energy within @xmath54 as @xmath55 where @xmath56 is the entropy density and @xmath57 is the mass - averaged velocity . applying the reynolds transport theorem and the gauss divergence theorem , we deduce that @xmath58 @xmath59 and @xmath60 in presence of a fluid velocity field , the mass transfer in fluids takes place through the convection in addition to the diffusion of each component .
thus , the mass balance law for component @xmath5 gives us @xmath61 where @xmath62 is the diffusion flux of component @xmath5 and its formulation will be discussed in the latter of this section .
multiplying by @xmath52 and summing them from @xmath63 to @xmath1 , we obtain the mass balance equation @xmath64 substituting into , we have @xmath65 where @xmath66 the work done by @xmath67 is expressed as @xmath68 cauchy s relation between face force @xmath67 and the stress tensor @xmath69 of component @xmath5 gives @xmath70 , and as a result , @xmath71 where @xmath72 is the unit normal vector towards the outside of @xmath54 .
we note that the other external forces including gravity force is ignored in this work , but the model derivations can be easily extended to the cases in the presence of external forces .
substituting , , and into , and taking into account the arbitrariness of @xmath54 , we obtain the entropy balance equation @xmath73 in the following subsections , we will derive the transport equation of helmholtz free energy density to reduce the entropy equation . from the definition of @xmath74
, we have @xmath75 using the relation and component mass balance equation , we derive the transport equation of helmholtz free energy density @xmath76 as @xmath77 the gradient contribution of helmholtz free energy density can be formulated as @xmath78 and @xmath79 combining - , we deduce the transport equation of the helmholtz free energy @xmath80 as @xmath81 taking into account the identity @xmath82 we can reformulate as @xmath83 substituting into , we reformulate the entropy equation as @xmath84 where @xmath85 is the second - order identity tensor .
we consider the fluid mixture in a closed system with the fixed total moles for each component .
let @xmath86 denote the domain with the fixed volume .
the natural boundary conditions can be formulated as @xmath87 where @xmath88 denotes a normal unit outward vector to the boundary @xmath89 .
integrating over the entire domain , we obtain the change of total entropy @xmath13 with time @xmath90 where @xmath91 . according to the second law of thermodynamics ,
the total entropy shall not decrease with time . using this principle
, we can determine the complete forms of multi - component two - phase flow model .
first , we consider an ideal reversible process to get the form of the reversible stress , and the reversibility implies that there exist no effects of viscosity and friction . in this case , the entropy shall be conserved , so the diffusions vanish , i.e. @xmath92 , and the total tress @xmath69 becomes equal to the reversible stress , denoted by @xmath93 , which must have the form @xmath94 the last term on the right - hand side of shall also be zero as @xmath95 for the realistic irreversible multi - component chemical systems , the driving force for diffusion of each component is the gradient of chemical potentials , so we express the diffusion flux for each component as @xcite @xmath96 where @xmath97 is the mobility .
reciprocal principle @xcite requires the symmetry of @xmath98 , and moreover , the second law of thermodynamics demands that @xmath98 is positive semidefinite or positive definite , i.e. @xmath99 which ensures the non - negativity of the first term on the right - hand side of . in general principle ,
@xmath100 may depend on the chemical potential gradients of other components except for component @xmath5 . in numerical tests of this paper ,
we take a special case of @xmath98 that is a diagonal positive definite matrix ; indeed , we use the following diffusion flux @xcite @xmath101 where @xmath102 is the diffusion coefficient of component @xmath5 . for the realistic viscous flow , the total stress can be split into two parts : reversible part ( i.e. @xmath93 given in ) and irreversible part ( denoted by @xmath103 ) ; that is , @xmath104 newtonian fluid theory suggests @xmath105 where @xmath106 is the volumetric viscosity and @xmath107 is the shear viscosity .
we assume that @xmath108 and @xmath109 .
so the second term on the right - hand side of is non - negative .
the non - negativity of the last term on the right - hand side of requires that @xmath110 substituting , and into , we obtain the complete momentum balance equation as @xmath111 which is also equivalent to @xmath112 if @xmath62 is taken such that @xmath113 , then is reduced into the form in @xcite , but for the general formulations of @xmath62 , the term @xmath114 or @xmath115 is essential to ensure the thermodynamical consistency @xcite . in summary , the proposed mathematical model of multi - component two - phase flow is formulated by a nonlinear and fully coupled system of equations including the mass conservation equation coupling with the diffusion flux and the chemical potential , and the momentum balance equation or coupling with the pressure formulation .
in this section , we will demonstrate the total ( free ) energy dissipation property of the proposed model , which seems not an obvious conclusion from its derivation , but it is required for thermodynamical consistency and it is essential to design the stable and efficient numerical methods in the next section . to do this ,
we first reformulate the momentum conservation equation by a simpler form .
this requires the following theorem regarding the relations between the gradients of the pressure and chemical potentials .
[ lempreschptleqnmultigrad ] the gradients of the pressure and chemical potentials have the following relation @xmath116 we recall the relation @xmath117 proved in , and then taking into account chemical potential formulation and pressure formulation , we obtain @xmath118 thus , is obtained . as a direct application of
, the momentum conservation equation is reformulated as @xmath119 which indicates that the fluid motion is indeed driven by the gradients of chemical potentials . applying the formulation of momentum conservation equation given in
, we can conveniently prove the total ( free ) energy dissipation property of the proposed model .
we use @xmath120 and @xmath121 to represent the @xmath122 , @xmath123 or @xmath124 inner product and norm respectively .
we define the helmholtz free energy and kinetic energy within the entire domain @xmath86 as @xmath125 the sum of the helmholtz free energy and kinetic energy is dissipated with time , i.e. @xmath126 taking into account the mass balance equation , we obtain @xmath127 multiplying both sides of by @xmath15 and integrating it over @xmath86 , we obtain @xmath128 applying the formulation of @xmath15 , we derive the summation of the first term on the left - hand side of as @xmath129 we denote @xmath130^t$ ] , and further define the norm @xmath131 using and , we have the summation of from @xmath63 to @xmath1 as @xmath132 we combine and @xmath133 where we have also used @xmath134 therefore , the energy dissipation can be obtained from and .
in this section , we will design an energy - dissipated semi - implicit time marching scheme for simulating the proposed multi - component fluid model .
the key difficulties result from the complication of helmholtz free energy density and fully coupling relations between molar densities and velocity .
we resolve these problems through very careful physical observations with a mathematical rigor .
the gradient contribution to the helmholtz free energy density is always convex with respect to molar densities , so it shall be treated implicitly .
the challenge is to deal with the bulk helmholtz free energy density @xmath76 . for the pure substance , the ideal and repulsion terms , i.e. @xmath23 and @xmath24 , result in convex contributions to the helmholtz free energy density @xcite . for the multi - component mixtures ( @xmath135 )
, we can prove the ideal contribution @xmath23 is still convex with respect to molar densities since its hessian matrix is a diagonal positive definite matrix .
however , the repulsion term @xmath24 for multi - component mixtures is never convex due to cross products of multiple components , and it can be easily verified for the binary mixtures through eigenvalue calculations ( we omit the details here ) .
for the pure substance , the attraction term @xmath25 is proved to result in a concave contribution to the helmholtz free energy density @xcite .
but it may be not true for multi - component mixture ; in fact , numerical tests show that the maximum eigenvalues of its hessian matrix may be slightly larger than zero in some cases ( not presented here ) . in order to construct a strict convex - concave splitting for the case of multi - component mixture , we impose two auxiliary terms on the bulk helmholtz free energy density .
one term is the additional ideal term for the reason that the ideal term is a good approximation of the behavior of many gases and always convex .
the other is a separate repulsion term , denoted by @xmath136 , which reduces the cross products of multiple components in the original repulsion term and is expressed as @xmath137 we note that @xmath138 in terms of its physical meaning , so @xmath139 is well defined .
we define the summation of the above two auxiliary terms as @xmath140 which has a diagonal positive definite hessian matrix @xmath141 we now state a convex - concave splitting of the bulk helmholtz free energy density based on the above auxiliary terms .
let us define a parameter @xmath142 , and then we reformulate the bulk helmholtz free energy density as @xmath143 where @xmath144 @xmath145 the chemical potentials can be reformulated accordingly @xmath146 we make some remarks on the convex - concave splitting given in - : + ( a ) the reason that @xmath24 is considered in the convex part is an observation in numerical tests that positive eigenvalues of @xmath24 are usually large , while its negative eigenvalues are only slightly less than zero .
in contrast , @xmath25 is included in the concave part because numerical tests show that its negative eigenvalues are usually far from zero , while its positive eigenvalues are only slightly larger than zero .
moreover , the attraction force shall result in a concave contribution from the point view of physics @xcite .
+ ( b ) because of strong convexity of @xmath147 , if we choose a sufficiently large @xmath148 , the strict convex - concave splitting can be achieved for the bulk helmholtz free energy density of multi - component mixture . but a very large value for @xmath148 is not necessary in practical computations , and in fact , we just take @xmath149 in our numerical tests , which is enough to gain the numerical convex - concave splitting .
consequently , we assume that there exists a suitable @xmath142 such that the convexity of @xmath150 and concavity of @xmath151 hold .
we consider a time interval @xmath152 $ ] , where @xmath153 .
we divide @xmath154 into @xmath155 subintervals @xmath156 $ ] , where @xmath157 and @xmath158 .
furthermore , we denote @xmath159 . for any scalar @xmath160 or vector @xmath161
, we denote by @xmath162 or @xmath163 its approximation at the time @xmath164 .
we now state the semi - implicit time marching scheme .
first , a semi - implicit time scheme is constructed for the molar density balance equation as @xmath165 where @xmath166 @xmath167 we note that the mobility coefficients @xmath168 generally depend on the molar densities , so we use @xmath169 to denote their values computed from molar densities @xmath170 . in the previous section ,
we have reformulated the momentum balance equation by the form coupling with the chemical potentials rather than pressure .
a semi - implicit time scheme for the momentum conservation equation is constructed as @xmath171 here , the viscosities @xmath107 and @xmath106 are assumed to be constant .
we now demonstrate that the proposed semi - implicit time scheme satisfies the dissipation property of the total ( free ) energy .
we define the discrete helmholtz free energy and discrete kinetic energy at the time @xmath172 as @xmath173 the sum of the discrete helmholtz free energy and discrete kinetic energy is dissipated with time steps , i.e. @xmath174 multiplying both sides of by @xmath175 and integrating it over @xmath86 , we obtain @xmath176 we define the bulk and gradient contributions of the discrete helmholtz free energy as @xmath177 the convexity and concavity yields @xmath178 using the symmetry of @xmath34 , we can estimate the gradient contribution of helmholtz free energy as @xmath179 since @xmath180 , we obtain from , and that @xmath181 we denote @xmath182^t$ ] , and define the norm of @xmath183 as @xmath184 substituting into yields @xmath185 multiplying both sides of by @xmath186 and and integrating it over @xmath86 , we obtain @xmath187 multiplying by @xmath52 and summing it from @xmath63 to @xmath1 , we obtain the mass balance equation @xmath188 using , we estimate @xmath189 substituting into yields @xmath190 noticing that @xmath191 we combine and and then obtain @xmath192 which yields the energy dissipation .
the above semi - implicit time schemes lead to a nonlinear systems of equations , and consequently , a linearized iterative method is required .
the newton s method is commonly used to solve such systems . here
, we present a mixed iterative method , in which we apply newton s linearization for @xmath193 and treat the rest terms through a simple linearization approach .
we use the superscript @xmath194 to represent the iteration step ; that is , @xmath195 and @xmath196 denote the approximations of @xmath197 and @xmath186 at the @xmath194th iteration .
the initial guess solutions are taken from the previous time step , i.e. @xmath198 and @xmath199 .
once the approximations @xmath195 and @xmath196 are obtained , the new approximations @xmath200 and @xmath201 are calculated as @xmath202 @xmath203 @xmath204 @xmath205 where @xmath206 stands for the hessian matrix of @xmath207 .
the convergence of this iterative method will be studied in the future work .
in this section , we carry out numerical tests regarding multi - component two - phase fluid flow problems and verify the effectiveness of the proposed method .
we consider the fluid flow of a binary hydrocarbon mixture , which is composed of methane ( ch@xmath208 ) and decane ( nc@xmath209 ) .
we take the volumetric viscosity and the shear viscosity as @xmath210pa@xmath211s , and we take the diffusion coefficients in as @xmath212m@xmath213/s . in all numerical examples , the spatial domain is a square domain with the length @xmath214 nm , and a uniform rectangular mesh with @xmath215 elements is applied on the domain .
we employ the cell - centered finite difference method and the upwind scheme to discretize the mass balance equation and the finite volume method on the staggered mesh @xcite for the momentum balance equation , and these spatial discretization schemes can be equivalent to special mixed finite element methods with quadrature rules @xcite .
the stop criterion of the nonlinear solver is that the 2-norm of the relative variation of molar density and velocity between the current and previous iterations is less than @xmath216 , and in many cases , about 2 - 3 nonlinear iterations are required to reach this criterion .
the maximum nonlinear iterations are also set to be not larger than 5 for preventing too many loops . in this example
, we simulate the dynamical evolution of a square shape fluid droplet , which is initially located in the center of the domain .
the temperature of the fluid keeps 320k constant .
the initial densities of liquid and gas fluids are computed by pr - eos under the pressure 160bar and temperature 320k .
the initial molar densities of methane and decane in gas phase are @xmath217mol / m@xmath218 and @xmath219mol / m@xmath218 respectively , while the initial molar densities of methane and decane in liquid phase are @xmath220mol / m@xmath218 and @xmath221mol / m@xmath218 respectively .
the time step size is taken as @xmath222s , and 45 time steps are simulated .
figure [ squarech4andnc10totalenergytemperature320](a ) clearly depicts the strict dissipation of the total ( free ) energy with time steps .
figure [ squarech4andnc10totalenergytemperature320](b ) is a zoom - in plot of figure [ squarech4andnc10totalenergytemperature320](a ) in the later time steps , and it shows that the total ( free ) energy remains to decrease all the time .
these results verify the effectiveness of the proposed method .
the initial molar density distributions for methane and decane are illustrated in figure [ squarech4andnc10molardensityofch4temperature320k](a ) and figure [ squarech4andnc10molardensityofnc10temperature320k](a ) respectively . in figure [ squarech4andnc10molardensityofch4temperature320k](b)(c ) and figure [ squarech4andnc10molardensityofnc10temperature320k](b)(c )
, we show the changes of each component molar density in the dynamical evolution .
it is obviously observed that the droplet turns to a circle from its initial square shape under the effect of driving force ( i.e. the gradient of chemical potential of this species ) . in figures
[ squarech4andnc10velocitytemperature320k ] , we illustrate the velocity field and magnitudes of both velocity components at the last time step .
we simulate the dynamical evolution of a bubble , which is ellipse shaped in the center of the domain at the initial moment .
the temperature of the fluid is constant at 330k .
the initial molar densities of methane and decane in gas phase are @xmath223mol / m@xmath218 and @xmath224mol / m@xmath218 respectively , while the initial molar densities of methane and decane in liquid phase are @xmath225mol / m@xmath218 and @xmath226mol / m@xmath218 respectively .
we set the time step size @xmath227s , and carry out the simulation for 100 time steps .
we illustrate the total ( free ) energy profiles with time steps in figure [ ellipsech4andnc10totalenergytemperature330](a ) , and further present a zoom - in plot of the later time steps in figure [ ellipsech4andnc10totalenergytemperature330](b ) .
it is still observed that the total ( free ) energy is dissipated with time steps . in figure [ ellipsech4andnc10molardensityofch4temperature330k ] and figure [ ellipsech4andnc10molardensityofnc10temperature330k ]
, we illustrate the initial molar density configurations for methane and decane and the molar density changes with time steps for each component .
moreover , in figures [ ellipsech4andnc10velocitytemperature330k ] , we illustrate the velocity fields and magnitudes of velocity components at different time steps .
we can observe from these results that at the @xmath228-direction the fluids flow towards to the center from the left and right sides , while at the @xmath229-direction , the fluids are flowing from the center to the bottom and top sides . as a result
, the bubble tends towards a circle from its original ellipse shape . in this example , there are two bubbles at the initial moment , and the initial molar densities of methane and decane in gas and liquid phases are the same to those in example 2 .
the temperature of the fluid keeps constant at 330k .
we simulate its dynamical evolution for 120 time steps with the time step size @xmath227s .
the total ( free ) energy dissipation profile with time steps is plotted in figure [ twoellipsech4andnc10totalenergytemperature330](a ) , and figure [ twoellipsech4andnc10totalenergytemperature330](b ) is a zoom - in plot in the later time steps .
the results in figure [ twoellipsech4andnc10totalenergytemperature330 ] confirm that the total energy always decays with time steps and consequently the proposed method is effective .
figure [ twoellipsech4andnc10molardensityofch4temperature330k ] and figure [ twoellipsech4andnc10molardensityofnc10temperature330k ] depict the initial molar density configurations and molar density changes with time steps for methane and decane . in figures
[ twoellipsech4andnc10velocitytemperature330k ] , we illustrate the velocity fields and magnitudes of velocity components at different time steps . it can be observed from these simulated results that in the presence of gradients of chemical potentials , the bubbles is first emerging with each other in the sequent dynamical evolution , and at the finial time , the merged bubbles gradually tend to a circle .
the nvt - based framework is a latest alternative setting that is preferred over the npt - based framework to model multiphase fluid flow .
based on the principles of the nvt - based framework , a mathematical model is proposed to describe the multi - component two - phase flow with partial miscibility .
we combine the first law of thermodynamics and the related physical relations to derive the entropy balance equations , and then we derive a transport equation of the helmholtz free energy density . furthermore , using the second law of thermodynamics , we derive a set of unified equations for both interfaces and bulk phases that can describe the partial miscibility of two fluids .
a feature of this model is that a term involving mass diffusions is naturally included in the momentum equation to ensure consistency with thermodynamics .
we prove a relation between the pressure gradient and the gradients of chemical potentials , which leads to a new formulation of the momentum conservation equation , showing that the gradients of chemical potentials become the primary driving force of the fluid motion .
the total ( free ) energy dissipation with time is proved for the proposed model .
an energy - dissipation numerical scheme is proposed based on a convex - concave splitting of helmholtz free energy density and a careful treatment of the coupling relations between molar densities and velocity with a mathematical rigor . finally
, we present numerical results to verify the effectiveness of the proposed method .
[ appendix ] we describe the detailed formulations of helmholtz free energy density @xmath21 in : @xmath230 @xmath231 @xmath232 where @xmath233 is the overall molar density , and @xmath234 is the universal gas constant . here ,
@xmath235 and @xmath236 are the energy parameter and the covolume , respectively , and these parameters can be calculated as functions of the mixture composition and the temperature .
we denote by @xmath237 and @xmath238 the @xmath5th component critical temperature and critical pressure , respectively . for the @xmath5th component ,
let the reduced temperature be @xmath239 .
the parameters @xmath240 and @xmath241 are calculated as @xmath242 ^ 2,~~~~b_{i}= 0.07780\frac{rt_{c_i}}{p_{c_i}}.\end{aligned}\ ] ] the coefficients @xmath243 are calculated by the following formulas @xmath244 @xmath245 where @xmath246 is the acentric factor .
we let the mole fraction of component @xmath5 be @xmath247 .
then @xmath248 and @xmath236 are calculated by @xmath249 where @xmath250 the given binary interaction coefficients for the energy parameters .
we now show that for bulk homogeneous fluids , the pr - eos formulation is equivalent to the pressure equation .
in fact , in this case , @xmath251 and the corresponding chemical potential of component @xmath5 can be calculated as @xmath252 and furthermore , we deduce that @xmath253 substituting the above equation into yields @xmath254 which is just the pr - eos formulation .
finally , we introduce the formulations of the influence parameters .
the pure component influence parameters @xmath255 are given by @xcite @xmath256,\end{aligned}\ ] ] where @xmath257 and @xmath258 are the coefficients correlated merely with the acentric factor @xmath246 of the @xmath5th component by the following relations @xmath259 the cross influence parameter is generally described as the modified geometric mean of the pure component influence parameters @xmath260 and @xmath261 by @xmath262 where the parameters @xmath263 are binary interaction coefficients chosen to satisfy the symmetry @xmath264 and @xmath265 . in the numerical tests of this work , we take @xmath266 for @xmath267 . from the above formulations of the influence parameters , it is generally assumed that the influence parameters rely on the temperature but independent of the component molar density .
k. bao , y. shi , s. sun , and x .- p .
wang . a finite element method for the numerical solution of the coupled cahn - hilliard and navier - stokes system for moving contact line problems . ,
231(24 ) : 80838099 , 2012 .
j. kou and s. sun .
unconditionally stable methods for simulating multi - component two - phase interface models with peng - robinson equation of state and various boundary conditions . , 291(1 ) : 158182 , 2016 .
|
a general diffuse interface model with a realistic equation of state ( e.g. peng - robinson equation of state ) is proposed to describe the multi - component two - phase fluid flow based on the principles of the nvt - based framework which is a latest alternative over the npt - based framework to model the realistic fluids .
the proposed model uses the helmholtz free energy rather than gibbs free energy in the npt - based framework .
different from the classical routines , we combine the first law of thermodynamics and related thermodynamical relations to derive the entropy balance equation , and then we derive a transport equation of the helmholtz free energy density .
furthermore , by using the second law of thermodynamics , we derive a set of unified equations for both interfaces and bulk phases that can describe the partial miscibility of two fluids . a relation between the pressure gradient and chemical potential gradients
is established , and this relation leads to a new formulation of the momentum balance equation , which demonstrates that chemical potential gradients become the primary driving force of fluid motion .
moreover , we prove that the proposed model satisfies the total ( free ) energy dissipation with time . for numerical simulation of the proposed model
, the key difficulties result from the strong nonlinearity of helmholtz free energy density and tight coupling relations between molar densities and velocity . to resolve these problems
, we propose a novel convex - concave splitting of helmholtz free energy density and deal well with the coupling relations between molar densities and velocity through very careful physical observations with a mathematical rigor .
we prove that the proposed numerical scheme can preserve the discrete ( free ) energy dissipation .
numerical tests are carried out to verify the effectiveness of the proposed method .
multi - component two - phase flow ; diffuse interface model ; partial miscibility ; energy dissipation ; realistic equation of state .
65n12 ; 76t10 ; 49s05
|
recently the space of closed curves on a surface attracted a lot of attention .
the interest was initiated by the work of v. arnold @xcite , who axiomatically defined invariants @xmath10 and @xmath11 of generic curves on @xmath12 . in order to define axiomatically this kind of invariants on an arbitrary surface
@xmath1 one has to know the fundamental group of the space @xmath13 of all immersed closed curves on @xmath1 .
however , as far as i know , this group is not calculated in the literature . in this paper
we explicitly calculate it .
the knowledge of its properties allowed me @xcite to generalize in a natural way arnold s invariants to the case of generic curves on an arbitrary surface .
( the results of the well known paper by s. smale @xcite , where he calculated homotopy groups of the space of all immersed closed curves with the fixed initial point and the velocity vector at it , are not sufficient for this purpose . )
when the work described in this paper was complete and submitted to mathematica scandinavica , i received a preprint of a. inshakov @xcite containing similar results obtained by him independently .
( later the preprint of inshakov was broken into two parts @xcite and @xcite . )
_ acknowledgments _
i am deeply grateful to oleg viro and tobias ekholm for many enlightening discussions .
a _ curve _ is a smooth immersion of an oriented circle @xmath14 into a ( smooth ) surface @xmath1 . for a surface @xmath1
we denote by @xmath0 the space of all curves on @xmath1 .
two curves @xmath15 and @xmath16 are said to be _ regularly homotopic _
, if there exists a homotopy @xmath17 such that @xmath18 , @xmath19 , and @xmath20 is an immersion for every @xmath21 .
this means that @xmath15 and @xmath16 are in the same connected component of @xmath0 .
two ( oriented ) curves with a tangency point , at which the velocity vectors of the two curves are pointing in the same direction , are said to be _ direct tangent _ to each other at this point . for a surface
@xmath1 we denote by @xmath22 the spherical tangent bundle of @xmath1 and by @xmath23 the corresponding locally trivial @xmath14-fibration . for a curve @xmath24 on @xmath1
we denote by @xmath25 its lifting to @xmath22 , which maps every point @xmath26 to the direction of the velocity vector of @xmath24 at @xmath27 .
we fix a point @xmath28 on @xmath14 .
then a curve @xmath24 represents an element of @xmath29 , and @xmath25 represents an element of @xmath30 .
when there is no ambiguity , we denote these two elements by @xmath24 and @xmath31 respectively . for orientable surfaces the group @xmath32 appears to be much simpler than for nonorientable surfaces .
[ pi1s2 ] let @xmath33 and let @xmath24 be a curve on @xmath8
. then @xmath34 .
[ pi1t2 ] let @xmath35 ( torus ) and let @xmath24 be a curve on @xmath36 . then @xmath37 .
[ pi1orient ] let @xmath38 be an orientable surface ( not necessarily compact ) and let @xmath24 be a curve on @xmath1 . if @xmath24 represents a homotopically nontrivial loop on @xmath1 , then @xmath39 . if @xmath24 represents a homotopically trivial loop on @xmath1 , then @xmath40 .
the proofs of theorems [ pi1s2 ] , [ pi1t2 ] and [ pi1orient ] are given in subsections [ pfpi1s2 ] , [ pfpi1t2 ] , and [ pi1arbitrary ] , respectively .
[ pi1rp2 ] let @xmath41 and let @xmath24 be a curve on @xmath9
. then @xmath42 .
[ pi1k ] let @xmath43 ( klein bottle ) and let @xmath24 be a curve on @xmath44 .
if @xmath24 represents an orientation preserving loop on @xmath44 , then @xmath45 , provided that @xmath46 in @xmath47 for some @xmath48 projecting to an orientation reversing loop on @xmath44 , and @xmath49 otherwise . .
if @xmath24 represents an orientation reversing loop on @xmath44 , then @xmath50 is isomorphic to @xmath51 .
the following construction will be needed for a description of @xmath52 for @xmath24 representing a homotopically nontrivial loop on @xmath1 and @xmath53 .
let @xmath53 be a surface ( not necessarily compact ) , and let @xmath24 be a curve on @xmath1 such that @xmath54 .
let @xmath55 be the homotopy class of an oriented fiber of the @xmath14-fibration @xmath23 .
one can show , that there exists a unique maximal abelian subgroup @xmath56 containing @xmath57 , and that this @xmath58 is isomorphic to @xmath51 ( see also proposition [ preissman ] ) .
let @xmath59 be its generator .
consider a curve @xmath60 direct tangent to @xmath24 at @xmath61 , which realizes @xmath62 .
one can show , that @xmath63 can be presented in the unique way as @xmath64 ( see also the proof of theorem [ pi1nonorient ] ) .
[ pi1nonorient ] let @xmath65 be a nonorientable surface ( not necessarily compact ) and let @xmath24 be a curve on @xmath1 . if @xmath24 represents an orientation reversing loop on @xmath1 , then @xmath66 .
if @xmath24 represents a homotopically nontrivial orientation preserving loop on @xmath1 then : \a ) @xmath39 , provided that @xmath60 is an orientation preserving loop on @xmath1 , or that @xmath60 is an orientation reversing loop and @xmath67 for some nonzero @xmath68 and @xmath69 .
\b ) @xmath70 , provided that @xmath60 is an orientation reversing loop and @xmath71 for some nonzero @xmath68 .
if @xmath24 represents a homotopically trivial loop , then : \a ) @xmath72 is isomorphic to the subgroup of @xmath73 consisting of all the elements , which project to orientation preserving loops on @xmath1 , provided that @xmath31 is a homotopically nontrivial loop in @xmath22 .
( this means , cf .
[ h - principle ] , that @xmath24 is not regularly homotopic to the figure eight curve . )
\b ) @xmath74 , provided that @xmath31 is a homotopically trivial loop in @xmath22 .
( this means , cf .
[ h - principle ] , that @xmath24 is regularly homotopic to the figure eight curve . )
the proofs of theorems [ pi1rp2 ] , [ pi1k ] and [ pi1nonorient ] are given in subsections [ pfpi1rp2 ] , [ pfpi1k ] and [ pi1arbitrary ] , respectively .
[ pin ] let @xmath1 be a surface ( not necessarily compact or orientable ) and let @xmath24 be a curve on @xmath1 . if @xmath1 is equal to @xmath8 or @xmath9 , then @xmath75 and @xmath76 , @xmath7 . if @xmath4 , then @xmath77 , @xmath78 .
for the proof of theorem [ pin ] see subsection [ pfpin ] .
[ commute ] let @xmath1 be a surface , let @xmath22 be its spherical tangent bundle and let @xmath79 be a point .
let @xmath80 be the class of an oriented ( in some way ) fiber of the @xmath14-fibration @xmath81 .
[ h - principle ] the parametric @xmath85-principle , see @xcite page 16 , implies that @xmath0 is weak homotopy equivalent to the space @xmath86 of free loops in @xmath22 .
the corresponding mapping @xmath87 sends an immersion @xmath88 to a loop @xmath89 by mapping a point @xmath90 to the point in @xmath22 corresponding to the velocity vector of @xmath24 at @xmath91 .
[ loopspacepi1 ] * relations between the homotopy groups of the space @xmath22 and of the space of free loops in @xmath22 .
* let @xmath92 be a point in @xmath22 .
we denote by @xmath93 the space of all loops in @xmath22 based at @xmath92 .
* proof of lemma [ g1 ] .
* let @xmath95 be the mapping described above . a proposition proved by v.l .
hansen @xcite says that : if @xmath104 is a topological space with @xmath105 , then @xmath106 is isomorphic to @xmath107 .
( here @xmath108 is the space of free loops in @xmath104 and @xmath109 is an element of @xmath108 . )
one can verify that @xmath110 for any surface @xmath1 .
thus , we get that @xmath101 is isomorphic to @xmath111 . from the proof of the hansen s proposition it follows that the isomorphism is induced by @xmath112 .
[ preissman ] let @xmath115 be a surface ( not necessarily compact or orientable ) and let @xmath116 be a nontrivial commutative subgroup of @xmath117 .
then @xmath116 is infinite cyclic and there exists a unique maximal infinite cyclic group @xmath118 such that @xmath119 . *
proof of lemma [ preissman ] . *
it is well known that any closed @xmath1 , other than @xmath120 , admits a hyperbolic metric of a constant negative curvature , which is induced from the universal covering of it by the hyperbolic plane @xmath121 .
the theorem by a. preissman ( see @xcite pp . 258 - 265 ) says that if @xmath122 is a compact riemannian manifold with a negative curvature , then any nontrivial abelian subgroup @xmath123 is isomorphic to @xmath51 .
thus if @xmath124 is closed , then any nontrivial commutative @xmath125 is infinite cyclic .
the proof of the preissman s theorem given in @xcite is based on the fact , that if @xmath126 are nontrivial commuting elements , then there exists a geodesic in @xmath127 ( the universal covering of @xmath122 ) , which is mapped to itself under the action of these elements considered as deck transformations on @xmath127 .
moreover , these transformations restricted to the geodesic act as translations .
this implies , that if @xmath124 is a closed surface , then there exists a unique maximal infinite cyclic @xmath118 such that @xmath119 .
this gives the proof of the lemma for closed @xmath1 .
consider @xmath44 as a quotient of a rectangle modulo the identification on its sides shown in figure [ klein1.fig ] .
we can assume that @xmath61 coincides with the image of a corner of the rectangle , and that @xmath24 and the side @xmath137 of the rectangle are direct tangent at this point .
let @xmath59 and @xmath85 be the curves such that : @xmath138 , @xmath139 and @xmath140 .
( here @xmath137 and @xmath141 are the elements of @xmath142 realized by the images of the sides of the rectangle used to construct @xmath44 , see figure [ klein1.fig ] . )
the second and the third relations in this presentation follow from and . to get the first relation one notes that the identity @xmath144 implies that @xmath145 for some @xmath146 . but @xmath147 commutes with @xmath148 , since they can be lifted to @xmath149 the fundamental group of which is abelian .
hence @xmath150 .
we are going to prove that the statement of theorem [ pi1nonorient ] is true for any orientable surface @xmath38 and any nonorientable @xmath53 .
( we will see that @xmath63 can be presented in the unique way as @xmath64 for any @xmath169 . )
[ xineq0 ] * proof of theorem [ pi1orient ] and theorem [ pi1nonorient ] in the case of @xmath170 . *
consider a subgroup @xmath116 of @xmath29 generated by @xmath24 .
it is an infinite cyclic group ( see [ preissman ] ) .
there is a unique ( see [ preissman ] ) maximal infinite cyclic group @xmath171 such that @xmath119 .
let @xmath59 be the generator of @xmath172 .
let @xmath60 be a curve direct tangent to @xmath24 at @xmath61 representing this @xmath59 .
corollary [ center ] says that @xmath50 is isomorphic to @xmath173 .
take @xmath174 .
since @xmath31 and @xmath175 commute in @xmath176 , we get that their images under the projection @xmath177 commute in @xmath178 .
lemma [ preissman ] implies that these projections are in the subgroup @xmath172 .
the kernel of the homomorphism @xmath179 is generated by @xmath180 , the homotopy class of an oriented fiber of the @xmath14-fibration @xmath23 .
this fact and identities and show that there exist unique @xmath181 such that @xmath182 and @xmath183 .
\b ) a group isomorphic to @xmath51 generated by @xmath187 , provided that @xmath60 is an orientation reversing loop , and @xmath68 is odd .
( this means that @xmath24 represents an orientation reversing loop on @xmath1 . )
note also , that in this case @xmath188 .
[ xieq0 ] * proof of theorem [ pi1orient ] and theorem [ pi1nonorient ] in the case of @xmath193 . * from the exact homotopy sequence of the @xmath14-fibration @xmath23 we get that @xmath194 is generated by @xmath180 , the homotopy class of the fiber . since @xmath195 we get that there exists a @xmath196 such that @xmath197 .
lemma [ split ] says that @xmath50 is isomorphic to @xmath198 . for @xmath162 identities and
imply that @xmath199 coincides with the set of elements of @xmath30 , which project to orientation preserving loops on @xmath1 .
this finishes the proof of the theorem for @xmath195 and @xmath200 .
* proof of lemma [ split ] * fix @xmath206 .
we construct a homomorphism @xmath207 such that @xmath208 .
after this the exactness of the sequence and the fact that higher homotopy groups are abelian imply the statement of the lemma .
take @xmath212 , which represents a given element of @xmath213 .
let @xmath214 be the mapping which is a lifting of @xmath215 under the covering @xmath216 .
fix an orientation of @xmath8 . then for every @xmath217 the orientation of a small neighborhood of @xmath218 induces an orientation of a small neighborhood of @xmath219 .
set the value of @xmath59 on the element of @xmath227 represented by @xmath215 to be the element of @xmath228 represented by @xmath229 .
a straightforward verification shows that this @xmath59 is the desired homomorphism from @xmath213 to @xmath228 .
one verifies that @xmath110 and @xmath230 , @xmath7 , for @xmath1 equal to @xmath8 or @xmath9 .
now lemma [ split ] , statement [ isomorphism ] and the weak homotopy equivalence given by the @xmath85-principle ( see [ h - principle ] ) imply the first statement of the theorem .
( note that @xmath231 . )
statement [ isomorphism ] says that @xmath232 .
one verifies that @xmath233 , @xmath234 for @xmath4 .
the exactness of sequence implies , that @xmath235 , @xmath78 . using the weak homotopy equivalence given by the @xmath85-principle we get the second statement of the theorem .
99999 v.i .
arnold , _ plane curves , their invariants , perestroikas and classifications _ ,
singularities and bifurcations ( v.i .
arnold , ed . )
21 , ( 1994 ) , pp.3991 m. do carmo , _ riemannian geometry _ , birkhuser , boston ( 1992 ) m. gromov , partial differential relations , springer - verlag berlin heidelberg ( 1986 ) v.l .
hansen , _ on the fundamental group of the mapping space _ , compositio mathematica , vol .
28 ( 1974 ) , pp 3336 a. v. inshakov , _ invarianty tipa @xmath236 gladkih krivyh na dvumernyh mnogoobrazijah _ , preprint 1997 a. v. inshakov , _ gomotopicheskie gruppy krivyh na dvumernyh mnogoobraziyah _ , uspehi mat .
nauk , vol .
53 , n 2 ( 1998 ) , pp .
147 - 148 a. v. inshakov , _ invarianty tipa @xmath236 gladkih krivyh na dvumernyh mnogoobrazijah _ , to appear in funct .
s. smale , _ regular curves on riemannian manifolds _ , trans .
87 , ( 1958 ) , pp .
492513 v. tchernov , arnold - type invariants of curves on surfaces and homotopy groups of the space of curves , preprint , dept . of math . ,
uppsala univ . ,
sweden , u.u.d.m .
|
we explicitly calculate the fundamental group of the space @xmath0 of all immersed closed curves on a surface @xmath1 .
it is shown that @xmath2 , @xmath3 , for @xmath4 .
it is also proved that @xmath5 , and @xmath6 , @xmath7 , for @xmath1 equal to @xmath8 or @xmath9 .
_ this paper will appear in math .
scand . , _ _ probably in vol * 86 * ( no
. 1 ) ( 2000 ) , to be issued in april 2000 _ by a surface we mean any smooth two - dimensional manifold .
|
graphic lambda calculus consists of a class of graphs endowed with moves between them .
it might be considered a visual language in the sense of erwig @xcite .
the name `` graphic lambda calculus '' comes from the fact that it can be used for representing terms and reductions from untyped lambda calculus .
it s main move is called `` graphic beta move '' for it s relation to the beta reduction in lambda calculus .
however , the graphic beta move can be applied outside the `` sector '' of untyped lambda calculus , and the graphic lambda calculus can be used for other purposes than the one of visual representing lambda calculus . for other visual , diagrammatic representation of lambda calculus see the vex language @xcite , or david keenan s @xcite .
the motivation for introducing graphic lambda calculus comes from the study of emergent algebras .
in fact , my goal is to build eventually a logic system which can be used for the formalization of certain `` computations '' in emergent algebras , which can be applied then for a discrete differential calculus which exists for metric spaces with dilations , comprising riemannian manifolds and sub - riemannian spaces with very low regularity .
emergent algebras are a generalization of quandles , namely an emergent algebra is a family of idempotent right quasigroups indexed by the elements of an abelian group , while quandles are self - distributive idempotent right quasigroups . tangle diagrams decorated by quandles or racks are a well known tool in knot theory @xcite @xcite .
it is notable to mention the work of kauffman @xcite , where the author uses knot diagrams for representing combinatory logic , thus untyped lambda calculus .
also meredith and snyder@xcite associate to any knot diagram a process in pi - calculus , is there any common ground between these three apparently separated field , namely differential calculus , logic and tangle diagrams ? as a first attempt for understanding this , i proposed @xmath0-scale calculus @xcite , which is a formalism which contains both untyped lambda calculus and emergent algebras .
also , in the paper @xcite i proposed a formalism of decorated tangle diagrams for emergent algebras and i called `` computing with space '' the various manipulations of these diagrams with geometric content .
nevertheless , in that paper i was not able to give a precise sense of the use of the word `` computing '' .
i speculated , by using analogies from studies of the visual system , especially the `` brain a geometry engine '' paradigm of koenderink @xcite , that , in order for the visual front end of the brain to reconstruct the visual space in the brain , there should be a kind of `` geometrical computation '' in the neural network of the brain akin to the manipulation of decorated tangle diagrams described in our paper .
i hope to convince the reader that graphic lambda calculus gives a rigorous answer to this question , being a formalism which contains , in a sense , lambda calculus , emergent algebras and tangle diagrams formalisms .
[ [ acknowledgement . ] ] acknowledgement .
+ + + + + + + + + + + + + + + + this work was supported by a grant of the romanian national authority for scientific research , cncs
uefiscdi , project number pn - ii - id - pce-2011 - 3 - 0383 .
an oriented graph is a pair @xmath1 , with @xmath2 the set of nodes and @xmath3 the set of edges .
let us denote by @xmath4 the map which associates to any node @xmath5 the set of adjacent edges @xmath6 .
in this paper we work with locally planar graphs with decorated nodes , i.e. we shall attach to a graph @xmath1 supplementary information : 1 . a function @xmath7 which associates to any node @xmath5 an element of the `` graphical alphabet '' @xmath8 ( see definition [ defalp ] ) , 2 .
a cyclic order of @xmath6 for any @xmath5 , which is equivalent to giving a local embedding of the node @xmath9 and edges adjacent to it into the plane .
we shall construct a set of locally planar graphs with decorated nodes , starting from a graphical alphabet of elementary graphs . on the set of graphs we shall define local transformations , or moves .
global moves or conditions will be then introduced .
the graphical alphabet contains the elementary graphs , or gates , denoted by @xmath0 , @xmath10 , @xmath11 , @xmath12 , and for any element @xmath13 of the commutative group @xmath14 , a graph denoted by @xmath15 . here
are the elements of the graphical alphabet : 1 .
@xmath0 graph , @xmath10 graph , 2 .
@xmath11 graph , @xmath15 graph , 3 .
@xmath12 graph . with the exception of the @xmath12 , all other elementary graphs have three edges .
the graph @xmath12 has only one edge .
[ defalp ] there are two types of `` fork '' graphs , the @xmath0 graph and the @xmath10 graph , and two types of `` join '' graphs , the @xmath11 graph and the @xmath15 graph .
further i briefly explain what are they supposed to represent and why they are needed in this graphic formalism .
the @xmath0 gate corresponds to the lambda abstraction operation from untyped lambda calculus .
this gate has one input ( the entry arrow ) and two outputs ( the exit arrows ) , therefore , at first view , it can not be a graphical representation of an operation . in untyped lambda calculus
the @xmath0 abstraction operation has two inputs , namely a variable name @xmath16 and a term @xmath8 , and one output , the term @xmath17 .
there is an algorithm , presented in section [ constru ] , which transforms a lambda calculus term into a graph made by elementary gates , such that to any lambda abstraction which appears in the term corresponds a @xmath0 gate .
the @xmath10 gate corresponds to a fan - out gate .
it is needed because the graphic lambda calculus described in this article does not have variable names .
@xmath10 gates appear in the process of elimination of variable names from lambda terms , in the algorithm previously mentioned .
another justification for the existence of two fork graphs is that they are subjected to different moves : the @xmath0 gate appears in the graphic beta move , together with the @xmath11 gate , while the @xmath10 gate appears in the fan - out moves .
thus , the @xmath0 and @xmath10 gates , even if they have the same topology , they are subjected to different moves , which in fact characterize their `` lambda abstraction''-ness and the `` fan - out''-ness of the respective gates . the alternative , which consists into using only one , generic , fork gate , leads to the identification , in a sense , of lambda abstraction with fan - out , which would be confusing .
the @xmath11 gate corresponds to the application operation from lambda calculus .
the algorithm from section [ constru ] associates a @xmath11 gate to any application operation used in a lambda calculus term .
the @xmath15 gate corresponds to an idempotent right quasigroup operation , which appears in emergent algebras , as an abstractization of the geometrical operation of taking a dilation ( of coefficient @xmath13 ) , based at a point and applied to another point .
as previously , the existence of two join gates , with the same topology , is justified by the fact that they appear in different moves .
[ [ the - set - graph . ] ] 1 . the set graph .
+ + + + + + + + + + + + + + + + + we construct the set of graphs @xmath18 over the graphical alphabet by grafting edges of a finite number of copies of the elements of the graphical alphabet .
@xmath18 is the set of graphs obtained by grafting edges of a finite number of copies of the elements of the graphical alphabet . during the grafting procedure ,
we start from a set of gates and we add , one by one , a finite number of gates , such that , at any step , any edge of any elementary graph is grafted on any other free edge ( i.e. not already grafted to other edge ) of the graph , with the condition that they have the same orientation . for any node of the graph , the local embedding into the plane is given by the element of the graphical alphabet which decorates it .
the set of free edges of a graph @xmath19 is named the set of leaves @xmath20 .
technically , one may imagine that we complete the graph @xmath19 by adding to the free extremity of any free edge a decorated node , called `` leaf '' , with decoration `` in '' or `` out '' , depending on the orientation of the respective free edge .
the set of leaves @xmath20 thus decomposes into a disjoint union @xmath21 of in or out leaves .
moreover , we admit into @xmath22 arrows without nodes , , called wires or lines , and loops ( without nodes from the elementary graphs , nor leaves ) graphs in @xmath18 can be disconnected .
any graph which is a finite reunion of lines , loops and assemblies of the elementary graphs is in @xmath22 .
[ [ local - moves . ] ] 2 . local moves .
+ + + + + + + + + + + + + + + these are transformations of graphs in @xmath18 which are local , in the sense that any of the moves apply to a limited part of a graph , keeping the rest of the graph unchanged .
we may define a local move as a rule of transformation of a graph into another of the following form .
first , a subgraph of a graph @xmath23 in @xmath18 is any collection of nodes and/or edges of @xmath23 .
it is not supposed that the mentioned subgraph must be in @xmath18 .
also , a collection of some edges of @xmath23 , without any node , count as a subgraph of @xmath23 .
thus , a subgraph of @xmath23 might be imagined as a subset of the reunion of nodes and edges of @xmath23 . for any natural number @xmath9 and any graph @xmath23 in @xmath18 , let @xmath24 be the collection of subgraphs @xmath25 of the graph @xmath23 which have the sum of the number of edges and nodes less than or equal to @xmath9 .
a local move has the following form : there is a number @xmath9 and a condition @xmath26 which is formulated in terms of graphs which have the sum of the number of edges and nodes less than or equal to @xmath9 , such that for any graph @xmath23 in @xmath18 and for any @xmath27 , if @xmath26 is true for @xmath25 then transform @xmath25 into @xmath28 , where @xmath28 is also a graph which have the sum of the number of edges and nodes less than or equal to @xmath9 .
graphically we may group the elements of the subgraph , subjected to the application of the local rule , into a region encircled with a dashed closed , simple curve .
the edges which cross the curve ( thus connecting the subgraph @xmath25 with the rest of the graph ) will be numbered clockwise .
the transformation will affect only the part of the graph which is inside the dashed curve ( inside meaning the bounded connected part of the plane which is bounded by the dashed curve ) and , after the transformation is performed , the edges of the transformed graph will connect to the graph outside the dashed curve by respecting the numbering of the edges which cross the dashed line .
however , the grouping of the elements of the subgraph has no intrinsic meaning in graphic lambda calculus .
it is just a visual help and it is not a part of the formalism . as a visual help ,
i shall use sometimes colors in the figures .
the colors , as well , do nt have any intrinsic meaning in the graphic lambda calculus .
[ [ graphic - beta - move . ] ] 2.1 .
graphic @xmath29 move .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + this is the most important move , inspired by the @xmath29-reduction from lambda calculus , see theorem [ lambdathm ] , part ( d ) .
the labels `` 1 , 2 , 3 , 4 '' are used only as guides for gluing correctly the new pattern , after removing the old one . as with the encircling dashed curve , they have no intrinsic meaning in graphic lambda calculus .
this `` sewing braids '' move will be used also in contexts outside of lambda calculus !
it is the most powerful move in this graphic calculus .
a primitive form of this move appears as the re - wiring move ( w1 ) ( section 3.3 , p. 20 and the last paragraph and figure from section 3.4 , p. 21 in @xcite ) .
an alternative notation for this move is the following : a move which looks very much alike the graphic beta move is the unzip operation from the formalism of knotted trivalent graphs , see for example the paper @xcite section 3 . in order to see this ,
let s draw again the graphic beta move , this time without labeling the arrows : the unzip operation acts only from left to right in the following figure .
remarkably , it acts on trivalent graphs ( but not oriented ) .
let us go back to the graphic beta move and remark that it does not depend on the particular embedding in the plane .
for example , the intersection of the `` 1,3 '' arrow with the `` 4,2 '' arrow is an artifact of the embedding , there is no node there .
intersections of arrows have no meaning , remember that we work with graphs which are locally planar , not globally planar .
the graphic beta move goes into both directions . in order to apply the move
, we may pick a pair of arrows and label them with `` 1,2,3,4 '' , such that , according to the orientation of the arrows , `` 1 '' points to `` 3 '' and `` 4 '' points to `` 2 '' , without any node or label between `` 1 '' and `` 3 '' and between `` 4 '' and `` 2 '' respectively . then , by a graphic beta move , we may replace the portions of the two arrows which are between `` 1 '' and `` 3 '' , respectively between `` 4 '' and `` 2 '' , by the pattern from the lhs of the figure .
the graphic beta move may be applied even to a single arrow , or to a loop . in the next figure
we see three applications of the graphic beta move .
they illustrate the need for considering loops and wires as members of @xmath22 .
also , we can apply in different ways a graphic beta move , to the same graph and in the same place , simply by using different labels `` 1 '' , ... `` 4 '' ( here @xmath8 , @xmath30 , @xmath26 , @xmath31 are graphs in @xmath18 ) : a particular case of the previous figure is yet another justification for having loops as elements in @xmath22 .
these two applications of the graphic beta move may be represented alternatively like this : [ [ co - assoc - move . ] ] 2.2 .
( co - assoc ) move .
+ + + + + + + + + + + + + + + + + + + + + this is the `` co - associativity '' move involving the @xmath10 graphs .
we think about the @xmath10 graph as corresponding to a fan - out gate . by using co - assoc moves
, we can move between any two binary trees formed only with @xmath32 gates , with the same number of output leaves .
[ [ co - comm - move . ] ] 2.3 .
( co - comm ) move .
+ + + + + + + + + + + + + + + + + + + + this is the `` co - commutativity '' move involving the @xmath10 gate .
it will be not used until the section [ secbraid ] concerning knot diagrams .
[ [ a - r1a - move . ] ] 2.3.a ( r1a ) move .
+ + + + + + + + + + + + + + + + + this move is imported from emergent algebras .
explanations are given in section [ semer ] .
it involves an @xmath10 graph and a @xmath15 graph , with @xmath33 .
[ [ b - r1b - move . ] ] 2.3.b ( r1b ) move .
+ + + + + + + + + + + + + + + + + the move r1b ( also related to emergent algebras ) is this : [ [ r2-move . ] ] 2.4 .
( r2 ) move .
+ + + + + + + + + + + + + + + this corresponds to the reidemeister ii move for emergent algebras .
it involves an @xmath10 graph and two other : a @xmath15 and a @xmath34 graph , with @xmath35 .
this move appears in section 3.4 , p. 21 @xcite , with the supplementary name `` triangle move '' .
[ [ ext2-move . ] ] 2.5 .
( ext2 ) move .
+ + + + + + + + + + + + + + + + + this corresponds to the rule ( ext2 ) from @xmath0-scale calculus , it expresses the fact that in emergent algebras the operation indexed with the neutral element @xmath36 of the group @xmath14 has the property @xmath37 .
[ [ local - pruning . ] ] 2.6 .
local pruning .
+ + + + + + + + + + + + + + + + + + + local pruning moves are local moves which eliminate `` dead '' edges .
notice that , unlike the previous moves , these are one - way ( you can eliminate dead edges , but not add them to graphs ) .
[ [ global - moves - or - conditions . ] ] global moves or conditions .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + global moves are those which are not local , either because the condition @xmath26 applies to parts of the graph which may have an arbitrary large sum or edges plus nodes , or because after the move the graph @xmath28 which replaces the graph @xmath25 has an arbitrary large sum or edges plus nodes .
[ [ ext1-move . ] ] 2.7 .
( ext1 ) move .
+ + + + + + + + + + + + + + + + + this corresponds to the rule ( ext1 ) from @xmath0-scale calculus , or to @xmath38-reduction in lambda calculus ( see theorem [ lambdathm ] , part ( e ) for details ) .
it involves a @xmath0 graph ( think about the @xmath0 abstraction operation in lambda calculus ) and a @xmath11 graph ( think about the application operation in lambda calculus ) .
the rule is : if there is no oriented path from `` 2 '' to `` 1 '' , then the following move can be performed .
[ [ global - fan - out - move . ] ] 2.8 .
( global fan - out ) move .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + this is a global move , because it consists in replacing ( under certain circumstances ) a graph by two copies of that graph . the rule is : if a graph in @xmath19 has a @xmath10 bottleneck , that is if we can find a sub - graph @xmath39 connected to the rest of the graph @xmath23 only through a @xmath10 gate , then we can perform the move explained in the next figure , from the left to the right .
conversely , if in the graph @xmath23 we can find two identical subgraphs ( denoted by @xmath8 ) , which are in @xmath18 , which have no edge connecting one with another and which are connected to the rest of @xmath23 only through one edge , as in the rhs of the figure , then we can perform the move from the right to the left .
remark that ( global fan - out ) trivially implies ( co - comm ) .
( as an local rule alternative to the global fan - out , we might consider the following .
fix a number @xmath40 and consider only graphs @xmath41 which have at most @xmath40 ( nodes + arrows ) .
the @xmath40 local fan - out move is the same as the global fan - out move , only it applies only to such graphs @xmath41 .
this local fan - out move does not imply co - comm . ) [ [ global - pruning . ] ] 2.9 . global pruning .
+ + + + + + + + + + + + + + + + + + + + this a global move which eliminates `` dead '' edges .
the rule is : if a graph in @xmath19 has a @xmath12 ending , that is if we can find a sub - graph @xmath39 connected only to a @xmath12 gate , with no edges connecting to the rest of @xmath23 , then we can erase this graph and the respective @xmath12 gate .
the global pruning may be needed because of the @xmath0 gates , which can not be removed only by local pruning .
[ [ elimination - of - loops . ] ] 2.10 .
elimination of loops .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + it is possible that , after using a local or global move , we obtain a graph with an arrow which closes itself , without being connected to any node .
here is an example , concerning the application of the graphic @xmath29 move .
we may erase any such loop , or add one .
[ [ lambdagraphs . ] ] @xmath0graphs .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + the edges of an elementary graph @xmath0 can be numbered unambiguously , clockwise , by 1 , 2 , 3 , such that 1 is the number of the entrant edge .
a graph @xmath19 is a @xmath0-graph , notation @xmath42 , if : 1 . it does not have @xmath15 gates , 2 . for any node @xmath0 any oriented path in @xmath23 starting at the edge 2 of this node can be completed to a path which either terminates in a graph @xmath12 , or else terminates at the edge 1 of this node .
the condition @xmath42 is global , in the sense that in order to decide if @xmath42 we have to examine parts of the graph which may have an arbitrary large sum or edges plus nodes .
here i show how to associate to a lambda term a graph in @xmath18 , then i use this to show that @xmath29-reduction in lambda calculus transforms into the @xmath29 rule for @xmath18 .
( thanks to morita yasuaki for some corrections . ) indeed , to any term @xmath43 ( where @xmath44 is the set of lambda terms over the variable set @xmath45 ) we associate its syntactic tree .
the syntactic tree of any lambda term is constructed by using two gates , one corresponding to the @xmath0 abstraction and the other corresponding to the application .
we draw syntactic trees with the leaves ( elements of @xmath45 ) at the bottom and the root at the top .
we shall use the following notation for the two gates : at the left is the gate for the @xmath0 abstraction and at the right is the gate for the application .
remark that these two gates are from the graphical alphabet of @xmath18 , but the syntactic tree is decorated : at the bottom we have leaves from @xmath45 . also , remark the peculiar orientation of the edge from the left ( in tree notation convention ) of the @xmath0 gate . for the moment
, this orientation is in contradiction with the implicit orientation ( from down - up ) of edges of the syntactic tree , but soon this matter will become clear . we shall remove all leaves decorations , with the price of introducing new gates , namely @xmath10 and @xmath12 gates .
this will be done in a sequence of steps , detailed further .
take the syntactic tree of @xmath43 , drawn with the mentioned conventions ( concerning gates and the positioning of leaves and root respectively ) .
we take as examples the following five lambda terms : @xmath46 , @xmath47 , @xmath48 , @xmath49 and @xmath50 . [
[ step-1 . ] ] step 1 .
+ + + + + + + elimination of bound variables , part i. any leaf of the tree is connected to the root by an unique path .
start from the leftmost leaf , perform the algorithm explained further , then go to the right and repeat until all leaves are exhausted .
we initialize also a list @xmath51 of bound variables .
take a leaf , say decorated with @xmath52 . to this leaf
is associated a word ( a list ) which is formed by the symbols of gates which are on the path which connects ( from the bottom - up ) the leaf with the root , together with information about which way , left ( l ) or right ( r ) , the path passes through the gates .
such a word is formed by the letters @xmath53 , @xmath54 , @xmath55 , @xmath56 .
if the first letter is @xmath53 then add to the list @xmath30 the pair @xmath57 formed by the variable name @xmath16 , and the associated word ( describing the path to follow from the respective leaf to the root ) . then pass to a new leaf .
else continue along the path to the roof .
if we arrive at a @xmath0 gate , this can happen only coming from the right leg of the @xmath0 gate , thus we can find only the letter @xmath54
. in such a case look at the variable @xmath58 which decorates the left leg of the same @xmath0 gate .
if @xmath59 then add to the syntactic tree a new edge , from @xmath58 to @xmath16 and proceed further along the path , else proceed further .
if the root is attained then pass to next leaf .
examples : the graphs associated to the mentioned lambda terms , together with the list of bound variables , are the following . 1 .
@xmath46 has @xmath60 , @xmath47 has @xmath61 , @xmath62 has @xmath63 .
+ 2 . @xmath49 has @xmath64 , @xmath50 has @xmath64
. + [ [ step-2 . ] ] step 2 .
+ + + + + + + elimination of bound variables , part ii .
we have now a list @xmath30 of bound variables .
if the list is empty then go to the next step .
else , do the following , starting from the first element of the list , until the list is finished .
an element , say @xmath57 , of the list , is either connected to other leaves by one or more edges added at step 1 , or not .
if is not connected then erase the variable name with the associated path @xmath65 and replace it by a @xmath12 gate .
if it is connected then erase it , replace it by a tree formed by @xmath10 gates , which starts at the place where the element of the list were before the erasure and stops at the leaves which were connected to @xmath16 .
erase all decorations which were joined to @xmath16 and also erase all edges which were added at step 1 to the leave @xmath16 from the list .
examples : after the step 2 , the graphs associated to the mentioned lambda terms are the following . 1 .
the graphs of @xmath46 , @xmath47 , @xmath62 are + 2 .
the graphs of @xmath49 , @xmath50 are + remark that at this step the necessity of having the peculiar orientation of the left leg of the @xmath0 gate becomes clear .
remark also that there may be more than one possible tree of gates @xmath10 , at each elimination of a bound variable ( in case a bound variable has at least tree occurrences ) .
one may use any tree of @xmath10 which is fit .
the problem of multiple possibilities is the reason of introducing the ( co - assoc ) move .
[ [ step-3 . ] ] step 3 .
+ + + + + + + we may still have leaves decorated by free variables .
starting from the left to the right , group them together in case some of them occur in multiple places , then replace the multiple occurrences of a free variable by a tree of @xmath10 gates with a free root , which ends exactly where the occurrences of the respective variable are .
again , there are multiple ways of doing this , but we may pass from one to another by a sequence of ( co - assoc ) moves .
examples : after the step 3 , all the graphs associated to the mentioned lambda terms , excepting the last one , are left unchanged .
the graph of the last term , changes . 1 .
as an illustration , i figure the graphs of @xmath49 , left unchanged by step 3 , and the graph of @xmath50 : + let @xmath66 $ ] be a transformation of a lambda term @xmath8 into a graph @xmath67 $ ] as described previously ( multiple transformations are possible because of the choice of @xmath10 trees ) .
then : 1 . for any term @xmath8
the graph @xmath67 $ ] is in @xmath68 , 2 .
if @xmath67'$ ] and @xmath67"$ ] are transformations of the term @xmath8 then we may pass from @xmath67'$ ] to @xmath67"$ ] by using a finite number ( exponential in the number of leaves of the syntactic tree of @xmath8 ) of ( co - assoc ) moves , 3 .
if @xmath30 is obtained from @xmath8 by @xmath69-conversion then we may pass from @xmath67 $ ] to @xmath70 $ ] by a finite sequence of ( co - assoc ) moves , 4 .
let @xmath71 be two terms and @xmath52 be a variable .
consider the terms @xmath17 and @xmath72 $ ] , where @xmath72 $ ] is the term obtained by substituting in @xmath8 the free occurrences of @xmath16 by @xmath30 .
we know that @xmath29 reduction in lambda calculus consists in passing from @xmath73 to @xmath72 $ ]
. then , by one @xmath29 move in @xmath18 applied to @xmath74 $ ] we pass to a graph which can be further transformed into one of @xmath72 $ ] , via ( global fan - out ) moves , ( co - assoc ) moves and pruning moves , 5 . with the notations from ( d ) , consider the terms @xmath8 and @xmath75 with @xmath76 ; then the @xmath38 reduction , consisting in passing from @xmath77 to @xmath8 , corresponds to the ext1 move applied to the graphs @xmath78 $ ] and @xmath67 $ ] .
[ lambdathm ] [ [ proof . ] ] proof .
+ + + + + + \(a ) we have to prove that for any node @xmath0 any oriented path in @xmath67 $ ] starting at the left exiting edge of this node can be completed to a path which either terminates in a graph @xmath12 , or else terminates at the entry peg of this node , but this is clear . indeed , either the bound variable ( of this @xmath0 node in the syntactic tree of @xmath8 ) is fresh , then the bound variable is replaced by a @xmath12 gate , or else , the bound variable is replaced by a tree of @xmath10 gates .
no matter which path we choose , we may complete it to a cycle passing by the said @xmath0 node .
\(b ) clear also , because the ( co - assoc ) move is designed for passing from a tree of @xmath10 gates to another tree with the same number of leaves .
\(c ) indeed , the names of bound variables of @xmath8 do not affect the construction of @xmath67 $ ] , therefore if @xmath30 is obtained by @xmath69-conversion of @xmath8 , then @xmath70 $ ] differs from @xmath67 $ ] only by the particular choice of trees of @xmath10 gates . but
this is solved by ( co - assoc ) moves .
\(d ) this is the surprising , maybe , part of the theorem .
there are two cases : @xmath16 is fresh for @xmath8 or not .
if @xmath16 is fresh for @xmath8 then in the graph @xmath79 $ ] the name variable @xmath16 is replaced by a @xmath12 gate . if not , then all the occurrences of @xmath16 in @xmath8 are connected by a @xmath10 tree with root at the left peg of the @xmath0 gate where @xmath16 appears as a bound variable . in the case
when @xmath16 is not fresh for @xmath8 , we see in the lhs of the figure the graph @xmath74 $ ] ( with a remanent decoration of `` x '' ) .
we perform a graphic ( @xmath29 ) move and we obtain the graph from the right .
this graph can be transformed into a graph of @xmath72 $ ] via ( global fan - out ) and ( co - assoc ) moves .
the case when @xmath16 is fresh for @xmath8 is figured next .
we see that the graph obtained by performing the graphic ( @xmath29 ) move is the union of the graph of @xmath8 and the graph of @xmath30 with a @xmath12 gate added at the root . by pruning we are left with the graph of @xmath8 , which is consistent to the fact that when @xmath16 is fresh for @xmath8 then @xmath73 transforms by @xmath29 reduction into @xmath8 .
\(e ) in the next figure we see at the lhs the graph @xmath80 $ ] and at the rhs the graph @xmath67 $ ] .
the red asterisk marks the arrow which appears in the construction @xmath80 $ ] from the variable @xmath16 , taking into account the hypothesis @xmath76 .
we have a pattern where we can apply the ext1 move and we obtain @xmath67 $ ] , as claimed .
@xmath81 as an example , let us manipulate the graph of @xmath49 : we can pass from the lhs figure to the rhs figure by using a graphic ( @xmath29 ) move .
conversely , we can pass from the rhs figure to the lhs figure by using a ( global fan - out ) move .
these manipulations correspond to the well known fact that @xmath82 is left unchanged after @xmath29 reduction : let @xmath83 , then @xmath84 .
[ [ s - k - and - i - combinators - in - graph . ] ] @xmath85 , @xmath86 and @xmath87 combinators in @xmath18 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the combinators @xmath46 , @xmath88 and @xmath62 have the following correspondents in @xmath18 , denoted by the same letters : \(a ) by one graphic ( @xmath29 ) move @xmath89 transforms into @xmath8 , for any @xmath39 with one output .
\(b ) by two graphic ( @xmath29 ) moves , followed by a global pruning , for any @xmath90 with one output , the graph @xmath91 transforms into @xmath8 .
\(c ) by five graphic ( @xmath29 ) moves , followed by one local pruning move , the graph @xmath92 transforms into @xmath87 .
\(d ) by three graphic ( @xmath29 ) moves followed by a ( global fan - out ) move , for any @xmath93 with one output , the graph @xmath94 transforms into the graph @xmath95 .
[ pcombi ] [ [ proof.-1 ] ] proof .
+ + + + + + the proof of ( b ) is given in the next figure .
the proof of ( c ) is given in the following figure .
\(a ) and ( d ) are left to the interested reader .
the manipulations of graphs presented in this section can be applied for graphs which represent lambda terms .
however , they can also be applied for graphs which do not represent lambda terms .
[ [ fixed - points . ] ] fixed points .
+ + + + + + + + + + + + + let s start with a graph @xmath39 , which has one distinguished input and one distinguished output.i represent this as follows . for any graph @xmath96 with one output ,
we denote by @xmath97 the graph obtained by grafting the output of @xmath96 to the input of @xmath98 .
i want to find @xmath96 such that @xmath99 , where @xmath100 means any finite sequence of moves in graphic lambda calculus .
i call such a graph @xmath30 a fixed point of @xmath8
. the solution of this problem is the same as in usual lambda calculus .
we start from the following succession of moves : this is very close to the solution , we only need a small modification : [ [ grafting - application - or - abstraction ] ] grafting , application or abstraction ?
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + if the @xmath8 , @xmath30 from the previous paragraph were representing lambda terms , then the natural operation between them is not grafting , but the application . or , in graphic lambda calculus the application it s represented by an elementary graph , therefore @xmath101 ( seen as the term in lambda calculus which is obtained as the application of @xmath8 to @xmath30 ) is not represented as a grafting of the output of @xmath30 to the input of @xmath8 .
we can easily transform grafting into the application operation .
suppose that @xmath8 and @xmath30 are graphs representing lambda terms , more precisely suppose that @xmath8 is representing a term ( denoted by @xmath8 too ) and it s input represents a free variable @xmath16 of the term @xmath8 .
then the grafting of @xmath30 to @xmath8 is the term @xmath72 $ ] and the graph from the right is representing @xmath102 , therefore both graphs are representing terms from lambda calculus .
we can transform grafting into something else : this has no meaning in lambda calculus , but excepting the orientation of one of the arrows of the graph from the right , it looks like if the abstraction gate ( the @xmath0 gate ) plays the role of an application operation . [
[ zippers - and - combinators - as - half - zippers . ] ] zippers and combinators as half - zippers .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + let s take @xmath103 a natural number and let s consider the following graph in @xmath104 , called the n - zipper : at the left is the n - zipper graph ; at the right is a notation for it , or a `` macro '' .
the zipper graph is interesting because it allows to perform ( nontrivial ) graphic beta moves in a fixed order . in the following picture
is figured in red the place where the first graphic beta move is applied . in terms of zipper notation
this graphic beta move has the following appearance : we see that a n - zipper transforms into a ( n-1)-zipper plus an arrow .
we may repeat this move , as long as we can .
this procedure defines a `` zipper move '' : we may see the 1-zipper move as the graphic beta move , which transforms the 1-zipper into two arrows .
the combinator @xmath105 satisfies the relation @xmath106 . in the next figure
it is shown that @xmath107 ( figured in green ) , when applied to @xmath8 , is just a half of the 1-zipper , with an arrow added ( figured in blue ) . by opening the zipper
we obtain @xmath108 , as it should .
the combinator @xmath109 satisfies @xmath110 . in the next figure
the combinator @xmath111 ( in green ) appears as half of the 2-zipper , with one arrow and one termination gate added ( in blue ) .
after opening the zipper we obtain a pair made by @xmath108 and @xmath112 which gets the termination gate on top of it .
a global pruning move sends @xmath30 to the trash bin .
finally , the combinator @xmath113 satisfies @xmath114 .
the combinator @xmath115 ( in green ) appears to be made by half of the 3-zipper , with some arrows and also with a `` diamond '' added ( all in blue ) .
interestingly , the diamond looks alike the ones from the emergent algebra sector , definition [ defemers ]
. expressed with the help of zippers , the relation @xmath116 appears like this .
[ [ lists - and - currying . ] ] lists and currying .
+ + + + + + + + + + + + + + + + + + + with the help of zippers , we may enhance the procedure of turning grafting into the application operation .
we have a graph @xmath117 which has one output and several inputs .
we use an n - zipper in order to clip the inputs with the output .
this graph is , in fact , the following one .
we may interpret the graph inside the green dotted rectangle as the currying of @xmath108 , let s call him @xmath118 .
this graph has only one output and no inputs .
the graph inside the red dotted rectangle is almost a list .
we shall transform it into a list by using again a zipper and one graphic beta move . [ [ packing - arrows . ] ] packing arrows .
+ + + + + + + + + + + + + + + we may pack several arrows into one .
i describe first the case of two arrows .
we start from the following sequence of three graphic beta moves . with words ,
this figure means : we pack the 1 , 2 , entries into a list , we pass it trough one arrow then we unpack the list into the outputs 3 , 4 .
this packing - unpacking trick may be used of course for more than a pair of arrows , in obvious ways , therefore it is not a restriction of generality to write only about two arrows .
we may apply the trick to a pair of graphs @xmath108 and @xmath112 , which are connected by a pair of arrows , like in the following figure . with the added packing and unpacking triples of gates , the graphs @xmath108 , @xmath112 are interacting only by the intermediary of one arrow .
in particular , we may use this trick for the elementary gates of abstraction and application , transforming them into graphs with one input and one output , like this : if we use the elementary gates transformed into graphs with one input and one output , the graphic beta move becomes this almost algebraic , 1d rule : with such procedures , we may transform any graph in @xmath18 into a 1d string of graphs , consisting of transformed elementary graphs and packers and un - packers of arrows , which could be used , in principle , for transforming graphic lambda calculus into a text programming language .
emergent algebras @xcite @xcite are a distillation of differential calculus in metric spaces with dilations @xcite .
this class of metric spaces contain the `` classical '' riemannian manifolds , as well as fractal like spaces as carnot groups or , more general , sub - riemannian or carnot - carathodory spaces , bellache @xcite , gromov @xcite , endowed with an intrinsic differential calculus based on some variant of the pansu derivative @xcite . in @xcite section 4
`` binary decorated trees and dilatations '' , i propose a formalism for making easy various calculations with dilation structures .
this formalism works with moves acting on binary decorated trees , with the leaves decorated with elements of a metric space .
here is an example of the formalism .
the moves are ( with same names as those used in graphic lambda calculus , see the explanation further ) : define the following graph ( and think about it as being the graphical representation of an operation @xmath119 with respect to the basepoint @xmath16 ) : then , in the binary trees formalism i can prove , by using the moves r1a , r2a , the following `` approximate '' associativity relation ( it is approximate because there appear a basepoint which is different from @xmath16 , but which , in the geometric context of spaces with dilations , is close to @xmath16 ) : it was puzzling that in fact the formalism worked without needing to know which metric space is used .
moreover , reasoning with moves acting on binary trees gave proofs of generalizations of results from sub - riemannian geometry , while classical proofs involve elaborate calculations with pseudo - differential operators . at a close inspection
it looked like somewhere in the background there is an abstract nonsense machine which is just applied to the particular case of sub - riemannian spaces . in this paper
i shall take the following pure algebraic definition of an emergent algebra ( compare with definition 5.1 @xcite ) , which is a stronger version of the definition 4.2 @xcite of a @xmath14 idempotent right quasigroup , in the sense that here i define a @xmath14 idempotent quasigroup .
let @xmath14 be a commutative group with neutral element denoted by @xmath36 and operation denoted multiplicatively .
a @xmath14 idempotent quasigroup is a set @xmath45 endowed with a family of operations @xmath120 , indexed by @xmath33 , such that : 1 . for any @xmath121
the pair @xmath122 is an idempotent quasigroup , i.e. for any @xmath123 the equations @xmath124 and @xmath125 have unique solutions and moreover @xmath126 for any @xmath52 , 2 .
the operation @xmath127 is trivial : for any @xmath128 we have @xmath129 , 3 . for any @xmath130 and
any @xmath35 we have : @xmath131 .
[ defirq ] here are some examples of @xmath14 idempotent quasigroups .
[ [ example-1 . ] ] example 1 .
+ + + + + + + + + + real ( or complex ) vector spaces : let @xmath45 be a real ( complex ) vector space , @xmath132 ( or @xmath133 ) , with multiplication as operation .
we define , for any @xmath33 the following quasigroup operation : @xmath134 .
these operations give to @xmath45 the structure of a @xmath14 idempotent quasigroup .
notice that @xmath135 is the dilation based at @xmath16 , of coefficient @xmath13 , applied to @xmath58 .
[ [ example-2 . ] ] example 2 .
+ + + + + + + + + + contractible groups : let @xmath23 be a group endowed with a group morphism @xmath136 . let @xmath137 with the operation of addition of integers ( thus we may adapt definition [ defirq ] to this example by using `` @xmath138 '' instead of `` @xmath139 '' and `` @xmath140 '' instead of `` @xmath36 '' as the name of the neutral element of @xmath14 ) . for any @xmath141 let @xmath142 .
this a @xmath143 idempotent quasigroup .
the most interesting case is the one when @xmath144 is an uniformly contractive automorphism of a topological group @xmath23 .
the structure of these groups is an active exploration area , see for example @xcite and the bibliography therein .
a fundamental result here is siebert @xcite , which gives a characterization of topological connected contractive locally compact groups as being nilpotent lie groups endowed with a one parameter family of dilations , i.e. almost carnot groups .
[ [ example-3 . ] ] example 3 .
+ + + + + + + + + + a group with an invertible self - mapping @xmath136 such that @xmath145 , where @xmath146 is the identity of the group @xmath23 .
it looks like example 2 but it shows that there is no need for @xmath144 to be a group morphism . [ [ local - versions . ] ] local versions .
+ + + + + + + + + + + + + + + we may accept that there is a way ( definitely needing care to well formulate , but intuitively clear ) to define a local version of the notion of a @xmath14 idempotent quasigroup .
with such a definition , for example , a convex subset of a real vector space gives a local @xmath147 idempotent quasigroup ( as in example 1 ) and a neighbourhood of the identity of a topological group @xmath23 , with an identity preserving , locally defined invertible self map ( as in example 3 ) gives a @xmath143 local idempotent quasigroup .
[ [ example-4 . ] ] example 4 .
+ + + + + + + + + + a particular case of example 3 is a lie group @xmath23 with the operations defined for any @xmath148 by @xmath149 .
[ [ example-5 . ] ] example 5 .
+ + + + + + + + + + a less symmetric example is the one of @xmath45 being a riemannian manifold , with associated operations defined for any @xmath148 by @xmath150 , where @xmath151 is the metric exponential .
[ [ example-6 . ] ] example 6 .
+ + + + + + + + + + more generally , any metric space with dilations is a local idempotent ( right ) quasigroup .
[ [ example-7 . ] ] example 7 .
+ + + + + + + + + + one parameter deformations of quandles .
a quandle is a self - distributive quasigroup .
take now a one - parameter family of quandles ( indexed by @xmath33 ) which satisfies moreover points 2 . and 3 .
from definition [ defirq ] .
what is interesting about this example is that quandles appear as decorations of knot diagrams @xcite @xcite , which are preserved by the reidemeister moves ( more on this in the section [ secbraid ] ) . at closer examination , examples 1 , 2 are particular cases of one parameter quandle deformations !
i define now the operations of approximate sum and approximate difference associated to a @xmath14 idempotent quasigroup . for any @xmath152 we give the following names to several combinations of operations of emergent algebras : 1 .
the approximate sum operation is @xmath153 @xmath154 , 2 .
the approximate difference operation is @xmath155 , 3 .
the approximate inverse operation is @xmath156 .
let s see what the approximate sum operation is , for example 1 .
@xmath157 it is clear that , as @xmath13 converges to @xmath140 , this becomes the operation of addition in the vector space with @xmath16 as neutral element , so it might be said that is the operation of addition of vectors in the tangent space at @xmath16 , where @xmath16 is seen as an element of the affine space constructed over the vector space from example 1 .
this is a general phenomenon , which becomes really interesting in non - commutative situations , i.e. when applied to examples from the end of the provided list .
these approximate operations have many algebraic properties which can be found by the abstract nonsense of manipulating binary trees .
another construction which can be done in emergent algebras is the one of taking finite differences ( at a high level of generality , not bonded to vector spaces ) .
let @xmath158 be a function ( from @xmath45 to itself , for simplicity ) .
the finite difference function associated to @xmath8 , with respect to the emergent algebra over @xmath45 , at a point @xmath52 is the following . @xmath159
[ defpan ] for example 1 , the finite difference has the expression : @xmath160 which is a finite difference indeed . in more generality , for example 2
this definition leads to the pansu derivative @xcite .
finite differences as defined here behave like discrete versions of derivatives . again , the proofs consist in manipulating well chosen binary trees .
all this can be formalized in graphic lambda calculus , thus transforming the proofs into computations inside graphic lambda calculus .
i shall not insist more on this , with the exception of describing the emergent algebra sector of graphic lambda calculus . for any @xmath161 ,
the following graphs in @xmath104 are introduced : 1 . the approximate sum graph @xmath162 1 .
the approximate difference graph @xmath163 1 . the approximate inverse graph @xmath164 [ defemers ] let @xmath108 be a set of symbols @xmath165 .
( these symbols will play the role of scale parameters going to @xmath166 . ) with @xmath108 and with the abelian group @xmath167 we construct a larger abelian group , call it @xmath168 , which is generated by @xmath108 and by @xmath167 .
now we introduce the emergent algebra sector ( over the set @xmath108 ) .
@xmath169 is the subset of @xmath104 ( over the group @xmath168 ) which is generated by the following list of gates :
1 . arrows and loops , 2 .
@xmath170 gate and the gates @xmath171 for any @xmath161 , 3 .
the approximate sum gate @xmath172 and the approximate difference gate @xmath173 , for any @xmath174 , with the operations of linking output to input arrows and with the following list of moves : 1 .
fan - out moves 2 .
emergent algebra moves for the group @xmath168 , 3 .
< pruning moves .
the set @xmath169 with the given list of moves is called the emergent algebra sector over the set @xmath108 .
[ defemerse ] the approximate inverse is not included into the list of generating gates .
that is because we can prove easily that for any @xmath174 we have @xmath175 .
( if @xmath161 then we trivially have @xmath176 because it is constructed from emergent algebra gates decorated by elements in @xmath167 , which are on the list of generating gates . ) here is the proof : we start with the approximate difference @xmath173 and with an @xmath170 gate and we arrive to the approximate inverse @xmath177 by a sequence of moves , as follows .
we proved the following relation for emergent algebras : @xmath178 .
this relation appears as a computation in graphic lambda calculus . as for the finite differences
, we may proceed as this .
a graph @xmath39 , with one input and one output distinguished , is computable with respect to the group @xmath179 if the following graph can be transformed by the moves from graphic lambda calculus into a graph which is made by assembling : 1 .
graphs from @xmath180 , 2 .
gates @xmath0 , @xmath11 and @xmath12 .
[ defemercomp ] it would be interesting to mix the emergent algebra sector with the lambda calculus sector ( in a sense this is already suggested in definition [ defemercomp ] ) . at first view
, it seems that the emergent algebra gates @xmath181 are operations which are added to the lambda calculus operations , the latter being more basic than the former .
i think this is not the case . in @xcite theorem 3.4 , in the formalism of lambda - scale calculus ( graphic lambda calculus is a visual variant of this ) , i show on the contrary that the emergent algebra gates could be applied to lambda terms and the result is a collection , or hierarchy of lambda calculi , organized into an emergent algebra structure .
this is surprising , at least for the author , because the initial goal of introducing lambda - scale calculus was to mimic lambda calculus with emergent algebra operations .
in this section we discuss about tangle diagrams and graphic lambda calculus . an oriented tangle is a collection of wired in 3d space , more precisely it is an embedding of a oriented one dimensional manifold in 3d space .
two tangles are the same up to topological deformation of the 3d space .
an oriented tangle diagram is , visually , a projection of a tangle , in general position , on a plane .
more specifically , an oriented tangle diagram is a globally planar oriented graph with 4-valent nodes which represent crossings of wires ( as seen in the projection ) , along with supplementary information about which wire passes over the respective crossing .
a locally planar tangle diagram is an oriented graph which satisfies the previous description , with the exception that it is only locally planar .
visually , a locally planar tangle diagram looks like an ordinary one , excepting that there may be crossings of edges of the graph which are not tangle crossings ( i.e. nodes of the graph ) . the purpose of this section is to show that we can `` simulate '' tangle diagrams with graphic lambda calculus .
this can be expressed more precisely in two ways .
the first way is that we can define `` crossing macros '' in graphic lambda calculus , which are certain graphs which play the role of crossings in a tangle diagram ( i.e. we can express the reidemeister moves , described further , as compositions of moves from graphic lambda calculus between such graphs ) . the second way is to say that to any tangle diagram we can associate a graph in @xmath18 such that to any reidemeister move is associated a certain composition of moves from graphic lambda calculus .
meredith ad snyder @xcite achieve this goal with the pi - calculus instead of graphic lambda calculus .
kauffman , in the second part of @xcite , associates tangle diagrams to combinators and writes about `` knotlogic '' .
two tangles are the same , up to topological equivalence , if and only if any tangle diagram of one tangle can be transformed by a finite sequence of reidemeister moves into a tangle diagram of the second tangle .
the oriented reidemeister moves are the following ( i shall use the same names as polyak @xcite , but with the letter @xmath182 replaced by the letter @xmath183 ) : in section [ semer ] , example 7 , it is mentioned that there is a connection between tangle diagrams and emergent algebras , via the notion of a quandle .
quandles are self - distributive idempotent quasigroups , which were invented as decorations of the arrows of a tangle diagram , which are invariant with respect to the reidemeister moves .
let us define the emergent algebra crossing macros .
( we can choose to neglect the @xmath13 decorations of the crossings , or , on the contrary , we can choose to do like in definition [ defemerse ] of the emergent algebra sector , namely to add a set @xmath8 to the group @xmath14 and use even more nuanced decorations for the crossings . ) in @xcite , sections 3 - 6 are dedicated to the use of these crossings for exploring emergent algebras and spaces with dilations .
all constructions and reasonings from there can be put into the graphic lambda calculus formalism . here
i shall explain only some introductory facts .
let us associate to any locally planar tangle diagram @xmath184 a graph in @xmath185 \in graph$ ] , called the translation of @xmath184 , which is obtained by replacing the crossings with the emergent crossing macros ( for a fixed @xmath13 ) . also , to any reidemeister move we associate it s translation in graphic lambda calculus , consisting in a local move between the translations of the lhs and rhs tangles which appear in the respective move .
( note : these translations are not added to the moves which define graphic lambda calculus . ) the translations of all oriented reidemeister moves of type 1 and 2 can be realized as sequences of the following moves from graphic lambda calculus : emergent algebra moves ( r1a , r1b , r2 , ext2 ) , fan - out moves ( i.e. co - comm , co - assoc , global fan - out ) and pruning moves .
more precisely the translations of the reidemeister moves r1a , r1b are , respectively , the graphic lambda calculus moves r1a , r1b , modulo fan - out moves .
moreover , all translations of reidemeister moves of type 2 can be expressed in graphic lambda calculus with the move r2 , fan - out and pruning moves .
[ temercross ] the fact that the reidemeister moves of type 3 are not true for ( the algebraic version of ) the emergent algebras , i.e. that the translations of those can not be expressed as a sequence of moves from graphic lambda calculus , is a feature of the formalism and not a weakness .
this is explained in detail in sections 5 , 6 @xcite , but unfortunately at the moment of the writing that article the graphic lambda calculus was not available .
it is an interesting goal the one of expressing the constructions from the mentioned sections as statements about the computability in the sense of definition [ defemercomp ] of the translations of certain tangle diagrams . as a justification for this point of view ,
let us remark that all tangle diagrams which appear in the reidemeister moves of type 3 have translations which are related to the approximate difference or approximate sum graphs from definition [ defemers ] .
for example , let s take the translation of the graph from the rhs of the move r3d and call it @xmath31 .
this graph has three inputs and three outputs .
let s then consider a graph formed by grafting three graphs @xmath8 , @xmath30 , @xmath26 at the inputs of @xmath31 , such that @xmath8 , @xmath30 , @xmath26 are not otherwise connected
. then we can perform the following sequence of moves .
the graph from the left lower side is formed by an approximate difference , a @xmath15 gate and several @xmath10 gates .
therefore , if @xmath8 , @xmath30 , @xmath26 are computable in the sense of definition [ defemers ] then the initial graph ( the translation of the lhs of r3d with @xmath8 , @xmath30 , @xmath26 grafted at the inputs ) is computable too .
as previously , we define translations of ( locally planar ) tangle diagrams into graphs in @xmath18 .
the class of locally planar tangle diagrams is out in a one - to one correspondence with a class of graphs in @xmath18 , let us call this class @xmath186 .
we could proceed in the inverse direction , namely consider the class of graphs @xmath186 , along with the moves : graphic beta move and elimination of loops .
then we make the ( inverse ) translation of graphs in @xmath186 into locally planar tangle diagrams and the ( inverse ) translation of the graphic beta move and the elimination of loops .
the following proposition explains what we obtain .
the class of graphs @xmath186 is closed with respect to the application of the graphic beta move and of the elimination of loops .
the translations of the graphic beta and elimination of loops moves are the following splice 1 , 2 ( translation of the graphic beta move ) and loop 1 , 2 ( translation of the elimination of loops ) moves .
[ ptrans ] the proposition becomes obvious if we find the translation of the graphic beta move .
there is one translation for each crossing .
( likewise , there are two translations for elimination of loops , depending on the orientation of the loop which is added / erased . ) @xmath81 the following theorem clarifies which are the oriented reidemeister moves which can be expressed as sequences of graphic lambda calculus moves applied to graphs in @xmath186 . among these moves ,
some are more powerful than others , as witnessed by the following all the translations of the oriented reidemeister move into moves between graphs in @xmath186 , excepting r2c , r2d , r3a , r3h , can be realized as sequences of graphic beta moves and elimination of loops .
moreover , the translations of moves r2c , r2d , r3a , r3h are equivalent up to graphic beta moves and elimination of loops ( i.e. any of these moves , together with the graphic beta move and elimination of loops , generates the other moves from this list ) .
[ threidgen ] it is easy , but tedious , to verify that all the mentioned moves can be realized as sequences of splice and loop moves .
it is as well easy to verify that the moves r2c , r2d , r3a , r3h are equivalent up to splice and loop moves .
it is not obvious that the moves r2c , r2d , r3a , r3h ca nt be realized as a sequence of splice and loop moves . in order to do this , we prove that r2d ca nt be generated by splice and loop .
thanks are due to peter kravchuk for the idea of the proof , given in an answer to a question i asked on mathoverflow @xcite , where i described the moves splice and loop . to any locally planar tangle diagram a associate it s reduced diagram r(a ) , which is obtained by the following procedure : first use splice 1,2 from left to right for all crossings , then use loop 1,2 from right to left in order to eliminate all loops which are present at this stage .
notice that : 1 .
the order of application of the splice moves does not matter , because they are applied only once per crossing .
there is a finite number of splices , equal to the number of crossings .
define the bag of splices splice(a ) to be the set of splice moves applied .
the same is true for the order of eliminations of loops by loop 1 , 2 .
there is a finite number of loop eliminations , because the number of loops ( at this stage ) can not be bigger than the number of edges of the initial diagram .
define the bag of loops loop(a ) to be the set of all loops which are present after all splices are done .
apply a splice 1,2 move to the initial diagram a , from left to right , and get b. then splice(b ) is what is left in the bag splice(a ) after taking out the respective splice . also loop(b )
= loop(a ) because of the definition of bags of loops . therefore r(a ) = r(b ) .
apply a loop1 , 2 from left to right to a and get b. the new loop introduced in the diagram does not participate to any crossing ( therefore splice(a ) = splice(b ) ) , so we find it in the bag of loops of b , which is made by all the elements of loop(a ) and this new loop . therefore r(a )
same goes for loop1 , 2 applied from right to left .
finally , remark that the reduced diagram of the lhs of the move r2d is different than the reduced diagram of the rhs of the move r2d , therefore the move r2d can not be achieved with a sequence of splices and loops addition / elimination . @xmath81 m. buliga , braided spaces with dilations and sub - riemannian symmetric spaces , in : geometry .
exploratory workshop on differential geometry and its applications , eds .
d. andrica , s. moroianu , cluj - napoca 2011 , 21 - 35 , http://arxiv.org/abs/1005.5031[arxiv:1005.5031 ]
|
we introduce and study graphic lambda calculus , a visual language which can be used for representing untyped lambda calculus , but it can also be used for computations in emergent algebras or for representing reidemeister moves of locally planar tangle diagrams .
[ section ] [ theorem]proposition [ theorem]lemma [ theorem]corollary [ theorem]definition [ theorem]remark [ section ]
|
the study of vortices in planar chern - simons ( cs ) models has been pioneered in refs . @xcite . since then , a lot of investigations on chern - simons vortices have been done ; see , e.g. , dunne1,dunne2,sakai , schap . during the past years , however , theories with non - canonical kinetic term , named generalized of @xmath0-field models , have been intensively studied .
generically , their applications have been found in strong interaction physics , with the skyrme @xcite and skyrme - like models @xcite , and also in cosmology with the so - called @xmath0-essence models @xcite .
@xmath0 fields change the way the fields approach their vacuum values , allowing thereby , for instance , the existence of solitons which approach their vacuum values in a power - like instead of an exponential fashion and which therefore have a compact support @xcite .
also , @xmath0-theories allow to avoid derrick s theorem @xcite increasing the chances to find soliton solutions in symmetry - breaking models . by this way
, several @xmath0-topological defects were already studied by several authors babichev1,jin , sarangi , bmp , bglm , bazeia0,bazeia3,babichev2,babichev3 and the overall conclusion is that their properties can be quite different from the standard ones depending specifically of the choice made for the kinetic term .
the non - linear effects in @xmath0-theories make the equations of motion more difficult to solve and therefore we will focus only on bps vortex solutions which minimize the energy .
they can be found by minimizing the energy functional @xcite or equivalently by using the conservation law for the energy - momentum tensor combined with the boundary conditions that require finite energy for the vortex @xcite .
this method combined with supersymmetry arguments allows for the first order formalism developed in ref .
@xcite , used to obtain bps global @xmath0 defects in one dimension and whose linear stability is proved analytically . in particular , when considering perturbative corrections to the canonical kinetic term , the authors found linearly stable solitons which , as expected , do not differ physically from the standard kinks , as they have the same energy , even though their width and energy densities are different .
they also found kink solutions through a specific combination for the non - canonical kinetic term and the potential .
finally , the method was used to obtain topological compactons @xcite , e.g. , solitons which approach the vacuum values at finite distance , confirming the results of ref .
thus , an important conclusion of this work is that even being a first order formalism ,
it is suitable for the study of nonlinearities in the kinetic term . more recently , supersymmetric extensions of @xmath0-field models has been introduced @xcite .
the most important aim of the present work is to generalize to vortices the first order formalism developed in ref .
we take a @xmath1 abelian chern - simons model with a non - canonical kinetic term for the complex scalar field .
we apply the method of ref .
@xcite to obtain the first order equations of motion , and then search for vortex solutions choosing the usual rotationally symmetric _ ansatz _ for the scalar and gauge fields .
first , we check that all vortex solutions of the bogomolnyi equations , i.e. , bps vortices , are physical requiring that they minimize the action .
we then study analytically the bps vortex equations and present a vortex solution for a specific choice for the non - canonical kinetic term .
finally , we compare our results with the bps vortex solutions obtained in refs.@xcite .
we use standard conventions , taking a @xmath1 spacetime with a plus minus signature for the minkowski metric @xmath2 and using bold style for the spatial components of 3-vectors .
we take an extension of the vortex model suggested by refs.@xcite , which has the standard form @xmath3 where @xmath0 is a constant , @xmath4 is the complex higgs field and @xmath5 is its potential . also , @xmath6 and @xmath7 , with @xmath8 being the electric charge . here
we are using @xmath9 , and the electric and magnetic fields are given by @xmath10 and @xmath11 , respectively .
we modify this model by changing the canonical kinetic term of the scalar field , as described by the new lagrangian density @xmath12 where @xmath13 is , in principle , an arbitrary function of the complex scalar field .
note that the non - canonical term in the lagrangian density , in the limit @xmath14 , leads us back to the standard chern - simons model .
it is convenient for the study of vortices to write all the variables in dimensionless units . for
that we take @xmath15 , where @xmath16 is a mass scale of the model . also , we take the two parameters @xmath0 and @xmath8 as : @xmath17 and the electric charge @xmath18 . in this case , we get @xmath19 and we can write @xmath20 , with @xmath21 being the lagrangian density to be used from now on .
the equations of motion for the gauge fields are given by @xmath22 where @xmath23 is the current density given by @xmath24.\ ] ] the time and spatial components of eq .
( [ maria1 ] ) are , for static field configurations , @xmath25and @xmath26which show that the electric charge density is proportional to the magnetic field while the density current is perpendicular to the electric field .
this fact is important for the phenomenological applications of chern - simons theories as effective field theories for the quantum hall effect @xcite .
the equation of motion for the scalar field is given by @xmath27 with @xmath28\ ] ] where @xmath29 is the determinant of the metric .
these second order differential equations will be reduced to first order ones by using the method developed in @xcite . for that , we need to obtain the components of the energy - momentum tensor , given by @xmath30}{\partial g^{\lambda \rho } } , \ ] ] where @xmath31 excludes the chern - simons term of the lagrangian density , since it does not contribute to the energy - momentum tensor . for the vortex
they become @xmath32with @xmath33\ ] ] writing explicitly the components of the energy - momentum tensor one obtains @xmath34\end{aligned}\ ] ] where @xmath35 we get to the first order equations of motion @xmath36 where @xmath37 with @xmath38
we now look for vortex solutions of the eqs .
( [ cesb1 ] ) and ( [ cesb2 ] ) , i.e. , bps vortices , for which we take the _ ansatz _ @xmath39 \ , ( \widehat{r } ^j \ , / r ) \label{a2}\end{aligned}\ ] ] where @xmath40 and @xmath41 are the polar coordinates and @xmath42 is the vortex winding number . for simplicity
, we set @xmath43 from now on .
inside the core , i.e. , near the origin , the boundary conditions are : @xmath44 , @xmath45 and @xmath46 , while faraway from it the vortex fields approach the vacuum , i.e. , @xmath47 , @xmath48 and @xmath49 .
this means that the potential has to have spontaneous symmetry breaking , as usual . the magnetic field , with magnitude @xmath50is parallel to the magnetic momentum , @xmath51 , @xmath52 and the angular momentum , @xmath53 , @xmath54 .
they all vanish near the origin and faraway from it .
also note that the magnetic flux , @xmath55 , and the electric charge , @xmath56 are quantized according to @xmath57 substituting the _ ansatz _ eqs .
( [ a1 ] ) and ( [ a2 ] ) into eq .
( [ cesb1 ] ) and ( [ cesb2 ] ) we get the bps vortex equations given by @xmath58 which substituted into eq .
( [ x ] ) gives @xmath59 we note that solutions of the first order equations eqs .
( [ cesb0 ] ) and ( [ cesb40 ] ) do satisfy the second order equations of motion ( [ maria1 ] ) and ( [ sf3 ] ) . for further reference
we also need to write the eqs .
( [ gl0 ] ) and ( sf3 ) as @xmath60=-\frac{2\,e^2}{k}\,w\,g \label{eev } \\ & & \frac{d \left(k a_0\right)}{dr } = 2 \ , e\ , w\,a\ , \frac{g^{2}}{r } \label{tempgauge}\end{aligned}\ ] ] which in particular gives that @xmath61 and @xmath62 are not independent .
we also need to write the energy density , the spatial component of the current density , the magnetic and angular momenta which respectively are given by @xmath63 we note that the procedure used in this section , to get to the first order equations , started with the conditions @xmath64 .
this is motivated by ref .
@xcite , and it shows explicitly that the choice of the potential @xmath65 depends on the choice of @xmath66 .
thus , neither the potential nor the function @xmath67 used to generalize the chern - simons model are arbitrary functions anymore .
in this section we review the vortex solution of the standard chern - simons model that is recovered by setting @xmath14 .
we use the eq .
( eev ) to get @xmath68whose solution gives the potential of the standard chen - simons model , which is @xmath69 here we have adjusted the integrating constant according to the vortex boundary conditions .
the first order equations ( [ cesb0 ] ) and ( [ cesb40 ] ) then become @xmath70 which can be integrated numerically from the infinity up to the origin .
for this , we need to write down the asymptotic solutions at the infinity which are @xmath71 with @xmath72 the modified bessel functions , @xmath73 and @xmath74 a constant which can be adjusted to get the suitable boundary conditions at the origin .
also note that as @xmath75it comes that near the origin @xmath76 while faraway from it @xmath77 . in the figures , we set @xmath78 and plot the higgs potential , the higgs and gauge fields , and the electric and magnetic fields . for that we used the asymptotic solutions @xmath79 for further reference it is also necessary to write the asymptotic solutions for the energy density , magnetic and angular momenta given by @xmath80 \label{asymp10 } \\
\frac{1}{e}\frac{d \mu^z}{dr}(r\rightarrow \infty)&=&2\pi m c\,r^{2 }
k_1(mr ) \label{asymp11 } \\
\frac{d l^z}{dr}(r\rightarrow \infty)&=&-4\pi m^2 c^2\,r^{2 } k_1(mr)\ , k_0(mr)\text{. } \label{asymp12}\end{aligned}\ ] ]
in this section we give an example of a vortex solution for the model introduced in sec .
ii . in order to make a choice for @xmath62 we first note from eq .
( [ eev ] ) that if @xmath62 is not a constant it changes the position of the zeros of the potential and its maximum amplitude when compared with those for the standard chern - simons potential .
we choose @xmath62 such that the zeros of @xmath61 are the same as the zeros of the higgs potential of the standard chern - simons model .
a possible choice for @xmath62 is @xmath81 which , from eqs .
( [ eev ] ) and ( [ tempgauge ] ) , gives @xmath82 ^ 8 \\ ea_0 = -\frac{e^2}{k}\left[1-g^2\right]^3\text{.}\end{aligned}\ ] ] thus , the electric field is given by @xmath83 ^ 2}{r}\ ] ] while the first order equations ( [ cesb0 ] ) and ( [ cesb40 ] ) become @xmath84 ^ 5\text{.}\end{aligned}\ ] ] the energy density , the polar component of the current density , the magnetic and angular momenta are respectively given by @xmath85 ^ 2\,\left[\left(\frac{dg}{dr}\right)^{2}+\frac{e^4}{k^2}\,g^2\,\left[1-g^2\right]^6\right ] \\ \
j^{\theta}&=&6\ , e\,\frac{g^{2}\,a\,\left[1-g^2\right]^2}{r } \\
\mu^z & = & 6\pi \ , e\ , \int \ , dr\ , r \,a \,g^2\,\left[1-g^2\right]^2 \\ l^z & = & -\frac{\pi\,k}{e^2}\end{aligned}\ ] ] note that this particular choice for @xmath62 allows the existence of vortices .
in fact , the vacuum manifold of the higgs potential is a dot and a circle which are not simply connected and the energy density is localized ( see vilenkin ) . in particular the electric and magnetic field vanish near the origin and faraway from it .
note that faraway from the origin the vortex solution approaches the standard vortex solution and therefore the asymptotic solutions are also given by the eqs .
( [ asymp1 ] ) , ( [ asymp2 ] ) , and ( [ asymp3])-([asymp12 ] ) . also note that there is no divergence in any physical quantity .
all this can be seen from the plots in figs .
[ 1]-[6 ] , where we show and compare the generalized vortex solution with the one of the standard chern - simons model . for the standard and generalized models .
conventions as in fig1 . note that in our model the vacuum is reached for a larger distance than in the standard model
see fig.3 as well ] . ]
( black/ red - dotdashed lines for standard / our model ) .
this is also the figure for the polar current density @xmath86 in units of @xmath0 .
note that it is smaller and spread over a larger distance than in the standard model . ] .
note that it is smaller and spread over a larger distance than in the standard model .
the same result applies for the charge density @xmath87 .
conventions as in fig .
[ fig4 ] . ] . ]
in this work we have studied the presence of vortices in a generalized chern - simons model .
the idea is different from the recent work babichev1 , where the author has investigated vortices in the maxwell - higgs model , modified to accommodate generalized structure , with the kinetic term being changed to a function of it .
this study was done with the numerical integration of the equations of motion , and the modification there introduced has induced another mass scale . here , our objective was to generalize the chern - simons model in a way such that we could find first order differential equations . to do this
, we have changed the kinetic term @xmath88 to @xmath89 .
we suppose that this modification leads to an effective planar field theory somehow similar to the standard chern - simons model . despite the modification in the kinematic scalar field term
, we could write a first order framework and find vortices which are qualitatively similar to the vortices of the standard chern - simons model .
however , we could identify several properties of the bps vortices which are quantitatively different from the standard vortices , since the solutions can be thicker then the standard solutions .
these differences are shown in all the figures , where we depict distinct features of the vortices in both the generalized and standard chern - simons models .
when compared to the model investigated in @xcite , an important distinction which appears in our work is that the modification we have included does not introduce another mass scale in the system . to see this clearly ,
let us write the potential in eq .
( [ gpot ] ) in terms of dimensional quantities .
it writes @xmath90 where @xmath91 is the symmetry breaking parameter of the model . in this case , the mass scale @xmath16 which we had to include at the end of sec .
ii can be seen as @xmath92 , so we do not need an extra mass scale , which had to be included in @xcite .
we are now examining how to obtain first order equations in a more general model , modifying the kinetic scalar field term but including both the maxwell and the chern - simons terms .
also , we are studying the presence of vortices in a maxwell - higgs model with the k - field modification similar to the case investigated in @xcite .
preliminary results indicate the presence of compact vortices , e.g. , of vortices with the scalar and gauge fields getting to their vacuum values at finite distances from the origin . before ending the work ,
let us study the bogomolnyi decomposition of the energy of the static solutions of the first order equations found above . to make the calculation explicit , we rewrite the energy density ( [ csed2 ] ) in the form @xmath93this result
can be used to recover the standard case of the chern - simons model .
we make @xmath94 to get to @xmath95with the standard potential @xmath96which leads to the first order equations obtained in sec .
iii . on the other hand ,
if we take @xmath97 we get @xmath98where we used @xmath99 we point out that an integration of @xmath100 over all planar space can be identified with an integration of @xmath101 the same space .
this integration process gives the magnetic flux @xmath102 which is topologically invariant . in this way
, this result leads to the first order equations used above , so the corresponding solutions are in fact bps states , with the energy bound being @xmath103 where @xmath104 represents the flux of the magnetic field in the plane .
we note that the energy bound in the generalized model is the same of the energy bound of the standard chern - simons model .
we thank capes and cnpq , brazil and christoph adam and filipe correia for interesting comments .
c dos santos is partially financed by sfrh / bsab/925/2009 , fct grant no .
cern / fp/109306/2009 . and would like to thank the departamento de fisica usc for all their hospitality while doing this work .
|
we search for vortices in a generalized abelian chern - simons model with a non - standard kinetic term .
we illustrate our results plotting and comparing several features of the vortex solution of the generalized model with those of the vortex solution found in the standard chern - simons model .
|
conventional superconductors in their superconducting state are characterized by opening of the superconducting gap ( sg ) in the electronic density of states ( dos ) .
superconductivity appears when electrons bind into cooper pairs and condense with long - range order below the superconducting transition temperature @xmath2 .
cuprate superconductors , however , as one of the most intriguing characteristics in their normal state , show the unusual emergence of the pseudogap ( pg ) in the electronic excitation spectrum even above @xmath2 , which persists up to a temperature @xmath3 , the pg onset temperature .
it has been widely accepted that understanding the pg origin and the relation between the pg and the sg may lead to a key to finding the basic mechanism of high - temperature superconductivity,@xcite which is not fully resolved up to the present .
there are two schools of thought as to understanding the pg in the cuprate physics : one - gap and two - gap ones .
one - gap school regards the pg as the precursor of the sg , where thermal fluctuations destroy long - range order while maintaining gap - like features in the excitation spectra in a certain high - temperature range ( @xmath4 ) of the normal state .
thus , the pg in question is believed to bring about the partial depletion of the dos at the normal - state fermi surface@xcite , resulting in the fermi arcs.@xcite the other school interprets the pg , especially in the underdoped regime , in terms of two gaps ; a small sg revealed in the nodal regions and a large gap of different origin in the antinodal regions . in the two - gap model , the fermi arcs are believed to emerge due to long - range order that , however , is not associated with the superconducting order . above @xmath2
the sg may vanish , leaving the other long - range order connected to the fermi arcs , fermi surface nestings@xcite or fermi surface pockets@xcite .
recent raman and angle - resolved - photoemission - spectroscopy ( arpes ) measurements show the consequences of opening of two gaps in underdoped single - layer hgba@xmath0cuo@xmath5 superconductors and bilayer bi@xmath0sr@xmath0ca@xmath6y@xmath7cu@xmath0o@xmath8 superconductors , respectively.@xcite the surface tunneling studies on the pg behavior suggest that the pg can evolve into the sg in the norm of one gap.@xcite in interlayer tunneling measurements on densely stacked intrinsic josephson junctions ( ijjs)@xcite formed in the layered cuprates , however , it has been proposed that the sg vanishes at @xmath2 and the pg may exist both below and above @xmath2 .
this experimental observation from ijjs is claimed to provide the norm of two gaps , where the sg and the pg are considered to be of different origins.@xcite the interlayer tunneling reveals the intrinsic bulk tunneling properties between cuo@xmath0 superconducting layers , which can be an advantage of this scheme compared to other surface - sensitive spectroscopic methods .
recently , however , it has been suggested that the experimental observations of the ijjs could be affected by the self - heating generated in a high - bias region , which was caused by the poor thermal conductivity of bi@xmath0sr@xmath0cacu@xmath0o@xmath8 ( bi-2212 ) and other materials involved in the tunneling measurements.@xcite the zero - bias tunneling process in the @xmath9 axis is very sensitive to the electronic dos at the normal - state fermi surface . in particular ,
the zero - bias tunneling resistance @xmath10 in the bi-2212 is weighted by the tunneling of quasiparticles in the antinodal region of the fermi surface.@xcite thus , @xmath10 is expected to increase rapidly as the pg opens and the corresponding dos is partially depleted at the fermi surface . in this point of view , the onset temperature of the pg opening can be defined as the characteristic temperature @xmath11 , at which @xmath10 deviates from the @xmath12-linear temperature dependence in @xmath10 vs @xmath12 curves.@xcite recently , kawakami _
et al._,@xcite based on the temperature dependence of @xmath10 , have shown that both the electron- and the hole - doped cuprates have common spin - singlet correlations in forming the pg , both of which thus close in high magnetic fields .
the closing fields of the pg and the sg , however , show much different temperature dependencies from each other , which indicates that the two gaps are of separate origins .
difference in origins of the two gaps is in line with the coexistence of the superconducting state and the pg state observed by the interlayer tunneling spectroscopy ( its ) in hole - doped bi-2212 ijjs , which is represented by the sharp peak and the broad hump structure below @xmath2.@xcite relating the hump structure in the high - energy windows of the its to the formation of the pg was controversial up to the present , however , again because of possible self - heating in a high - bias region,@xcite although there have been many efforts to reduce the self - heating effect in the its , by adopting schemes such as reducing the junction area , reducing the number of stacked junctions , and adopting pulsed biasing.@xcite in this study , for an overdoped bi-2212 sample fabricated on an as - grown single crystal , we discriminate the pg onset temperature , defined by the appearance of the hump structure ( @xmath13 ) in the its while lowering temperature , from that obtained by the @xmath10 vs @xmath12 behavior ( @xmath11 ) . to obtain the interlayer tunneling characteristics that are essentially free from self - heating artifact
, we adopted the recently developed technique of heating - compensated its , where a large portion of the bias - induced self - heating was removed.@xcite in contrast to @xmath11@xmath14 190 k , the hump structure persisted up to temperatures much higher than commonly perceived .
we then numerically illustrate that , with significant self - heating , @xmath13 would reduce down to @xmath11 , which confirms the previous reports by others in the heating - dominated case.@xcite we thus suggest that the @xmath13 represents the genuine onset of the pg . as in earlier works@xcite , with decreasing temperature below @xmath2 , the hump voltages in our heating - free its increases , along with the increase of the superconducting gap size .
it turns out that this unusual temperature variation of the hump voltage below @xmath2 results from the combined tunneling contribution of the quasiprticles associated with both the sg and the pg in the electronic state .
since the antinodal tunneling is weighted , the behavior of this pg obtained in the its of our overdoped bi-2212 should be related to the electronic structure of the antinodal region in the first brillouin zone of the cuo@xmath0 layers .
m apart laterally from the sample stack .
( b ) the temperature discrepancies between the sample stack and thermometer stacks as a function of base temperature of the substrate holding copper block for various heating powers dissipated in the sample stack .
inset : the thermal conductivities of bi-2212 along the c - axis direction and au electrode material as a function of temperature . ] fig .
1(a ) illustrates the schematic configuration of the sample .
we fabricated an overdoped sample stack of bi-2212 ijjs with the lateral size of 3@xmath153 @xmath16m@xmath17 , sandwiched between two thin - film ( top ; 400 nm thick , bottom ; 100 nm thick ) au electrodes .
this structure , where the pedestal stack ( large stack of ijjs outside of but coupled to the stack of ijjs of interest ) in the usual mesa structure was eliminated , gave more uniform tunneling current distribution.@xcite the number of junctions contained in the stack , @xmath18 , was 19 .
the hole concentration , @xmath19= 0.19 , was determined using the @xmath9-axis superconducting transition temperature @xmath2=88.3 k by the empirical relation of @xmath2=95[1 - 82.6(@xmath19 - 0.16)@xmath17 ] and the resistance ratio of @xmath10 between @xmath2 and @xmath12=300 k.@xcite a thermometer stack of ijjs , with the lateral dimension of 3@xmath152 @xmath16m@xmath17 , was arranged less than 1 @xmath16 m from the sample stack through a 100-nm - thick bottom au electrode . for the its the whole probe with a sample inside the vacuum
can was cooled down to the liquid - helium bath temperature .
prior to the its measurements in a finite bias the sample was set at a higher working temperature by using a resistive heater coil wound on the substrate - holding copper block .
the temperature variation of the sample stack during its measurements was monitored _ in - situ _ by the change in the tunneling resistance of the thermometer stack taken in a constant bias current@xcite of @xmath21 @xmath16a . to maintain the thermometer stack at a given set temperature during its measurements we compensated the bias - induced heating by lowering the current level to the heater coil wound on the copper substrate holder block .
we repeated this heating - compensation scheme by using a computer - aided proportional - integral - derivative control of the thermometer stack incorporated with adjusting the heating - current level . by adopting this technique
we were able to maintain the temperature of the thermometer stack within about 0.2 k in the whole bias range of the its.@xcite however , there could still be a temperature difference , @xmath22 , between the sample stack and the thermometer stack due to the finite thermal conductance of the bottom au electrode , through which the heat generated in the sample stack flowed to the thermometer stack , and due to the thermal leakage of heating to the surroundings.@xcite we numerically estimated the temperature difference @xmath22 between the sample stack and the thermometer stack during its measurements .
the comsol multiphysics program was used to calculate the temperature profile in the sample.@xcite in the estimation , we referred to the geometry and arrangement of the sample used in the measurement [ fig .
namely , the common bottom au electrode was attached to the sapphire substrate ( 0.4 mm thick and 5@xmath155 mm@xmath17 in the lateral size ) using negative photoresist , which in turn was fixed on the heater - coil - wound copper block using ge varnish ( assumed to be about 1 @xmath16 m thick ) .
the top au electrodes in the sample and thermometer stacks were extended by up to 200 @xmath16 m by two ( 300-nm - thick and 3-@xmath16m - wide ) au stripes ( per each stack ) deposited on the substrate that was precoated with an about 1-@xmath16m - thick negative photoresist layer .
the end of each au extension was then connected to a gauge-@xmath2340 copper wire that was thermally anchored at the base copper - block temperature .
the bottom common electrodes were also extended in the same manner by three au stripes ; two in the sample - stack side and one in the thermometer - stack side .
the inset of fig .
1(b ) shows the temperature dependence of the thermal conductivities of au and bi-2212 along the @xmath9 axis,@xcite which played a crucial role in the heat flow through the sample .
the in - plane thermal conductivity of bi-2212 was assumed to be ten times higher than the @xmath9-axis thermal conductivity.@xcite the thermal conductivities for sapphire , negative photoresist , and ge varnish were assumed to be insensitive to the temperature variation and set to be 40 , 0.2 , and 0.2 w / m - k , respectively .
the heat generated at the sample stack was dissipated through both the top and bottom thermal channels , while the temperature of the thermometer stack was maintained at a fixed temperature during the heating - compensated its .
1(b ) shows the discrepancy of temperature between the the sample stack and the thermometer stack as a function of the copper - block base temperature , corresponding to the heating power at the sample stack of 0.27 , 0.70 , and 1.35 mw .
one notices that @xmath22 is governed by the temperature dependence of the thermal conductivity of the bottom au electrode : @xmath22 increases in the temperature range of 20 - 80 k because of reduction of the au thermal conductivity until it saturates at temperatures above 100 k. the heating power of 0.70 mw corresponds to the bias voltage that is high enough to observe the hump structure in the differential conductance in the inset of fig . 2 ( @xmath24=525 mv ) and in figs . 4 and 5(a )
( @xmath25=@xmath24/19=27.6 mv ) .
then the estimation indicates that the heating - compensated thermometry adopted in this study allowed accuracy of the thermometry within 3.5 k for the heating power of 0.70 mw at any base temperatures .
this is in remarkable contrast to the discrepancy of several tens of degrees,@xcite usually encountered without the heating compensation incorporated with the _ in - situ _ thermometry .
thus , the hump structure obtained from this heating - compensated its can be regarded to be almost heating - free .
[ = @xmath26 curves of the overdoped sample .
the straight line in the figure is a guide to the eye , showing linear @xmath10 above a characteristic temperature @xmath11 .
@xmath10 may increase rapidly as the pseudogap opens and the corresponding electron density of states is partially depleted at the fermi surface . in this point of view , @xmath11 is often defined as the onset temperature of pseudogap opening.@xcite the inset : interlayer tunneling spectra @xmath27 , showing hump structures in the normal state above @xmath2 . ]
2 shows the @xmath10 vs @xmath12 curve in the normal state of the sample , which was obtained by the inverse of zero - bias tunneling conductance @xmath28 of the electronic excitation spectra in the inset of fig .
2 . this curve suggests the pg onset temperature to be @xmath29190 k. the normal state of our overdoped sample in the inset of fig .
2 shows a distinct zero - bias depletion of electronic excitation spectra with the pg size for each @xmath12 denoted by a pair of vertical segments .
the onset temperature of the pg opening can also be determined by the appearance of the hump structure ( @xmath13 ) in the tunneling @xmath27 spectra , which can be obtained either by the scanning tunneling spectroscopy@xcite ( sts ) or by the its .
in contrast to a naive expectation and the existing observations,@xcite however , the depletion of the dos near zero bias is evident even far above @xmath11 in @xmath10 vs @xmath12 curves .
the spectral depletion around zero bias persists up to the maximum temperature examined , _
i.e. _ , 241 k. we observed similar behavior in another overdoped sample .
but , the deviation from the @xmath12-linear behavior in @xmath10 vs @xmath12 is supposed to become evident only when the dos is sufficiently depleted at the characteristic temperature @xmath11 further below the onset temperature of the hump structure opening .
this indicates that @xmath13 better represents the onset of pg opening than @xmath11 . on the other hand , outside the gapped region ( @xmath30 @xmath31450 mv ) in the inset of fig .
2 , @xmath10 monotonically increases with increasing temperature over the whole temperature range examined , which represents a metallic behavior .
this pg onset temperature defined by @xmath13 at least in the overdoped regime is in clear contrast to that determined by the angle - resolved photoemission spectroscopy ( arpes ) and by the sts,@xcite where @xmath3 , representing the pg in the one - gap picture , disappears or merges into the bell - shaped @xmath2 curve near the optimal doping point in the temperature - vs - doping - level phase diagram . on the other hand ,
@xmath13 becomes comparable to @xmath11 in the presence of significant self - heating .
based on the heating - compensated @xmath27 curves in the normal state of the inset of fig .
2 , we simulated the @xmath27 curves of a stack under the influence of the serious self - heating .
3(a ) displays @xmath10-vs-@xmath12 curves obtained from the @xmath27(@xmath24 ) spectra given in the inset of fig . 2 for varying biases from 0 to 360 mv ( or from 0 to 18.9 mv per junction ) . with increasing the bias voltage ,
the up - turn deviation from the @xmath12-linear @xmath10-vs-@xmath12 behavior gradually disappears .
the sample temperature increases by a bias power defined at a fixed voltage with a given heating ratio [ k / mw ] .
the heating ratio in a stack of ijjs is determined by the junction area and the number of junctions .
the new value of @xmath27 at an increased temperature due to self - heating for a finite bias voltage was traced through the @xmath10-vs-@xmath12 curves of the corresponding voltage in fig .
since current - voltage ( _ i - v _ ) characteristics had almost the same curvatures in the temperature range under investigation we assumed that the power at a fixed voltage remained almost the same in the temperature range . with increasing bias voltages .
( b ) and ( c ) temperature dependence of @xmath27 curves estimated based on the curves in ( a ) for finite self - heating with the heating ratio of 60 and 150k / mw , respectively .
the dashed curves in ( b ) and ( c ) illustrate the temperature variation from the base temperature of 140k and 225 k , respectively , by the bias - induced self - heating for the two values of the heating ratio . ]
3(b ) and 3(c ) show the calculated @xmath27 curves , which would be affected by self - heating with the heating ratio of 60 k / mw and 150 k / mw , respectively . figures also show the effective temperature of the sample as a function of the bias voltage for two base temperatures , @xmath32=140 k and 225 k. since the hump structures in these figures are weakened by the self - heating the voltage positions and the heights of the humps get smaller than the ones shown in the inset of fig .
2 . in fig .
3(c ) , a higher heating ratio than that in fig .
3(b ) makes the hump structure disappear at a lower temperature around @xmath12@xmath14210 k , which is close to @xmath11 .
this calculation clearly shows that the disappearance of the hump structure near @xmath11 is indeed due to self - heating.@xcite our nonlinear @xmath27 curves in the normal state are in contrast to the flat @xmath27 behavior modeled for the normal state in ref .
the hump structures for temperatures above @xmath14170 k as in the inset of fig .
2 , which lead to the local minimum of @xmath10 in the main panel , can not be explained , either , in terms of the self - heating model with a flat @xmath27 behavior .
therefore , the hump structure previously reported in the its@xcite should not have been solely from the self - heating effect but from the intrinsic depletion of the zero - bias electronic spectral weight , presumably affected by the self - heating . this pg behavior with high onset temperature , @xmath3 , in the overdoped regime , revealed by our heating - compensated its , has characteristics similar to that observed in the angle - integrated photoemission spectroscopy and in the electronic magnetic susceptibility @xmath33.@xcite this peculiar pg behavior displays a clear peak - dip - hump structure below but near @xmath2 , where the peak pertains to the superconducting coherence . for our overdoped sample as a function of bias voltage per junction ( @xmath25 ) at various @xmath12 .
inset : _ i - v _ and @xmath34 curves calculated using eq .
( 1 ) for @xmath35=1 ( dotted curves ) and @xmath35=@xmath36 ( solid curves ) at @xmath12=4.2 k. ] a series of the overall feature of heating - compensated interlayer tunneling spectra @xmath34 of our overdoped sample , for varying @xmath12 , is displayed in fig .
4 , where the voltage is normalized by the number of junctions as @xmath25=@xmath37 . in the normal state above
@xmath2 the low - bias dos is smoothly depleted , revealing the pg . at a @xmath12 below @xmath2 a sharper peak ( the coherence peak )
develops inside the pg , constituting the peak - dip - hump structure . further lowering @xmath12 ,
the fast sharpening coherence peak with the growing sg size overwhelms the spectrum , leaving only the coherence peak apparent .
the pg with the hump becomes more conspicuous in underdoped samples ( not shown ) .
below @xmath2 , the tunneling spectra show the more u - shaped dos in the subgap region than the one observed previously.@xcite the fluctuating conductance at zero bias sufficiently below @xmath2 in fig . 4 was caused by the josephson pair tunneling .
curves for @xmath12=80 and 82.6 k in our overdoped sample .
( b ) the shape of imagined electron density of states of two identical superconducting electrodes of a junction , which contain both the superconducting coherence peak and the pg hump , with assumed gap sizes of @xmath38=20 mev and @xmath39=80 mev .
one electrode is assumed to be biased with @xmath24=100 mv .
inset of ( a ) : temperature dependence of voltage positions of the superconducting coherence peak ( @xmath40 ) and the pg hump ( @xmath41 ) , along with the corresponding sg energy ( @xmath38 ) and pg energy ( @xmath39 ) . ] in the following we discuss features of the superconducting gaps and pseudogaps and the interrelation between them one can observe in or deduce from the heating - free its results .
5(a ) shows @xmath34 curves at @xmath12=80 k and 82.6 k below @xmath2 of our overdoped sample .
the electronic spectra near and below @xmath2 show a peak - dip - hump structure .
as the coherence peak in the its starts to develop below @xmath2 the height and voltage position of the peak increases with decreasing temperature as illustrated in figs . 4 and 5(a ) .
this represents the sg edge .
the hump structure in the normal state in the inset of fig .
2 is connected to that below @xmath2 , which thus represents the pg state .
this observation is consistent with the recently reported raman response functions , which reveal a change of spectral weight from lower energies to higher ones when making a transition from the normal state into the superconducting state at the optimal doping and slightly overdoped levels .
in particular , the changes of the raman responses are more enhanced in the antinodal regime than in the nodal regime.@xcite the coexistence of the superconducting and pg states below @xmath2 reflects that the two gaps are of different origins.@xcite this is in contradiction to the characteristics of the one - gap concept observed in arpes and sts , where the pg is smoothly connected to the sg near @xmath2 in the varied doping ranges.@xcite the hump structure in the peak - dip - hump excitation spectral distribution provides valuable information on the interrelation between the pg and superconducting states .
the inset of fig .
5(a ) shows the temperature dependence of the superconducting coherence peak voltage ( @xmath40 ) and the pg hump voltage ( @xmath41 ) near @xmath2 .
since the tunneling occurs between two neighboring superconducting layers , the sg energy @xmath38 should be one half of @xmath42 .
in fact , the value of @xmath40 is seen to reduce rapidly along with the sg as the temperature approaches to @xmath2(=88 k ) from below .
however , it turns out that the pg energy @xmath39 determined in relation with @xmath43 , for the superconductor - insulator - superconductor ( sis ) tunneling near @xmath2,@xcite should be defined in a somewhat different way . for temperatures in the range of @xmath12@xmath44@xmath2 , where @xmath38 vanishes completely , @xmath39 is supposed to be simply @xmath43/2 .
but for @xmath12@xmath45@xmath2 the value of @xmath41 is affected by opening of the sg as well as the pg .
the differential conductance as a function of the voltage bias in the sis tunneling junction is given by@xcite @xmath46de,\end{aligned}\ ] ] where @xmath47 and @xmath48 are the electron dos of two identical superconducting layers with the @xmath49 symmetry , @xmath35 is the tunneling matrix and @xmath50 [ = tan@xmath51 is the azimuthal tunneling angle .
the angle - integrated electronic dos , @xmath52 for an assumed @xmath53 in the presence of both the superconducting and the pg states , is illustrated in fig .
5(b ) . here
, two different kinds of quasiparticles dressed by the superconducting state and the pg state are assumed to be accumulated near @xmath38=20 mev and @xmath39=80 mev , respectively .
quasiparticles fill all states below zero bias .
if the voltage bias , @xmath24 , is applied to the counter - electrode superconductor @xmath48 is shifted by @xmath54 along the energy axis . in this process ,
quasiparticles with an energy @xmath55 in the occupied region of @xmath47 tunnel to the unoccupied states of @xmath48 at the same energy .
this tunneling weight determines the conductance at a given voltage bias in the measurement .
especially , superconducting coherence peak at @xmath55=@xmath38 of @xmath47 dominantly determines the conductance shape as a function of voltage if the peak is sufficiently higher than the hump in the electronic dos .
the angle - integrated dos of the counterelectode , @xmath56 , shown in fig .
5(b ) is the case where the bias voltage @xmath24=100 mv corresponds to ( @xmath38+@xmath39)/e . here , the superconducting coherence peak and the pg hump filled with the quasiparticles of @xmath47
are respectively overlapped with the vacant pg hump and the superconducting coherence peak of @xmath48 . in this condition ,
the conductance is enhanced at the voltage bias corresponding to the hump voltage in the @xmath57 curves .
thus , below @xmath2 , @xmath43=@xmath38+@xmath39 rather than simply @xmath58 .
this inclusion of the influence of opening of the sg , @xmath38 , in @xmath41 _ explains the reason why @xmath41 increases along with @xmath40 below @xmath2 _ as seen in the inset of fig .
thus , care should be taken in extracting gap values from @xmath57 curves of an sis junction .
if the height of superconducting coherence peak is not much higher than that of the pg in the tunneling dos close to @xmath2 , the @xmath43 turns out to be positioned between @xmath59 and @xmath58 .
thus , the @xmath41 can be smoothly connected near @xmath2 as in the inset of fig .
5(a ) and also as reported previously.@xcite
if @xmath38 gets close in its value to @xmath39 at sufficiently low temperatures , one can not easily distinguish the hump position from the coherence peak position because @xmath38+@xmath39@xmath14@xmath60 .
this is the reason why no hump structure is visible below @xmath1480 k in the inset of fig .
@xcite the inset of fig .
5(a ) also displays the temperature dependence of the two characteristic gaps extracted using the above analysis . with decreasing temperature near and below @xmath2 , the pg increases abruptly .
this unusual behavior arises because the apparent hump structure in the tunneling spectra of a junction largely depends on which spectral weight , the sg or the pg , of an electrode is coupled to the pg spectral weight of the counterelectrode ; ( i ) for @xmath12@xmath44@xmath2 , the pg dos in one electrode is detected by the broad pseudogap dos in the counterelectrode , which makes @xmath39 lower than the expected position because of the broadness and ( ii ) @xmath12@xmath45@xmath2 , pg dos in one electrode is detected by the sharp superconducting peak in the opposite electrode , giving the value of @xmath39 close to the expectation .
this dos analysis indicates that the hump structure at temperature below @xmath2 is due to tunneling of quasiparticles associated with the sg ( pg ) in one electrode of a junction to a vacant quasiparticle state associated with the pg ( sg ) in the counter electrode . in this tunneling process of the quasiparticles ,
the key observation is that _ the quasiparticle tunneling constituting the hump structure is possible only if the sg and the pg arise from the combined electronic state _ of quasiparticles of two different characters in the same momentum space , _ i.e. _ , the antinodal region .
this picture implies that quasiparticles in a single state distribute either in the sg spectra or in the pg spectra , depending on external physical parameters such as temperature , magnetic field , doping , etc .
this inference is consistent with our earlier observation that , in a several - tesla @xmath9-axis magnetic field , the tunneling spectral weight in stacks of bi-2212 ijjs ( for both overdoped and underdoped ones ) redistributes from the superconducting coherence peak to the pg hump.@xcite but , the fact that the superconducting and the pg states are based on the single electronic structure composed of quasiparticles of two different characters is in contradiction to the earlier tunneling measurements with claiming that the peak - dip - hump features arise from a simple overlap of spectral functions of antibonding and bonding states , associated with the bilayer splitting@xcite . the anisotropic tunneling matrix element@xcite in the interlayer tunneling in the bi-2212 filters out the tunneling in the nodal region and weights the tunneling near the antinodal points on the momentum space in the brillouin zone . the pronounced u - shape in the measured tunneling @xmath27 curves of figs .
4 and 5(a ) , which is in contrast to the v - shape ones usually observed in the sts,@xcite is caused by this filtering .
the dotted curves in the inset of fig .
4 show the numerically obtained _ i - v _
( @xmath25 is the bias voltage per junction ) and the differential conductance curves using eq .
( 1 ) for a @xmath61-independent tunneling matrix element @xmath35=1,@xcite with @xmath38=20 mev and the quasiparticle scattering rate @xmath62= 0.05 mev for an assumed dos , @xmath63^{1/2}\}$ ] at @xmath12=4.2 k. the solid curves in the inset of fig .
4 correspond to the anisotropic tunneling matrix element @xmath64 , which is theoretically predicted for a crystal with the tetragonal symmetry as bi-2212.@xcite here , @xmath65 is the hopping constant .
this anisotropic @xmath35 reduces the low - energy quasiparticle tunneling near the nodal points and leads to the u - shaped tunneling conductance , while sharpening the coherence peak @xcite .
thus , the its in our heating - compensation scheme mainly shows the electronic state in the vicinity of the antinodal region and the pg formation is more closely related to the electronic state in this region.@xcite it is widely accepted that a fermi - surface nesting exists with a van hove singularity ( high dos with a flat band ) near the antinodal region,@xcite which is related to the formation of an antiferromagnetic order or orders like the spin - density wave and the charge - density wave .
this fermi surface nesting may be related to the downturn behavior of background spectra of the its@xcite in the inset of fig .
the fermi nesting near the antinodal region is reduced with increasing the doping in hole - doped cuprates because of change of the fermi surface topology with doping.@xcite the angle - integrated photoemission spectroscopy also showed that the binding energy of the pg corresponding to the flat - band position of the antinodal region , the so - called high - energy pseudogap , decreases with increasing doping.@xcite thus , one can expect that the pg energy scale and the pg onset temperature will decrease with increasing doping if the pg in the its is associated with the antinodal electronic state .
indeed , it has been reported that the pg onset temperature @xmath11 and the pg closing field @xmath66 observed in @xmath10 vs @xmath12 decrease with increasing doping.@xcite the low - energy pseudogap is believed to be a precursor of the superconducting state , while the high - energy pseudogap ( hepg ) is inferred to be of an antiferromagnetic order or orders like the spin - density wave and the charge - density wave . in the hole - doped cuprates
, the low - energy pseudogap tends to close near the optimal doping but the hepg persists even in the heavily overdoped regime .
features of the hepg related to the antinodal region have been observed in various experiments such as the electronic magnetic susceptibility,@xcite the knight shift,@xcite and the angle - integrated photoemission spectroscopy.@xcite the onset temperatures of the hepg over the doping values in these measurements were almost twice as high as those of the low - energy pseudogap .
especially , the onset temperatures of the hepg for @[email protected] were 260@xmath14270 k , close to our its results at the same doping value .
since the interlayer tunneling spectra mainly reveal the electronic state in the vicinity of the antinodal region the formation of the pseudogap is closely related to the electronic state in this region .
it has been widely accepted that the antinodal regions in cuprates are related to the formation of an antiferromagnetic order or orders like the spin - density wave and the charge - density wave .
it thus strongly indicates that our hump structure is highly likely to be related to the hepg .
in order to understand the nature of the pg , the interlayer tunneling spectroscopic characteristics of high-@xmath2 superconductors were investigated , while the self - heating was largely excluded using the heating - compensation technique incorporated with the _ in - situ _ thermometry . since the its is sensitive to the bias - induced self - heating an extreme care
should be taken , as in this study , to keep the sample temperature constant within a tolerance limit over the whole bias sweeping range .
but , as demonstrated by krasnov _
et al._,@xcite most of the essential observations on the pg feature remain valid even in the earlier its results .
thus , its , with a precautious measure taken to eliminate the self - heating , should provide a very useful experimental tools to investigate the electronic excitation spectrum of highly anisotropic materials containing naturally grown tunneling junctions . in this study
, it is found that the genuine pg behavior in the overdoped cuprates reveals the following characteristics .
defined by the appearance of the hump at @xmath13 in the differential tunneling conductance , the pg onset temperature @xmath3 reaches up to nearly room temperature , much higher than the estimation based on the tunneling resistive transition , @xmath11 . with significant self - heating
, however , numerical simulation shows that @xmath13 returns to @xmath11 .
this observation indicates that the hump structure in the tunneling differential conductance provides far more accurate determination of the pg onset temperature than the tunneling resistive transition . the hump voltage revealed in the its below @xmath2
is shown to follow the sg value , which is , in fact , additional confirmation that the hump structure in the its represents a genuine electronic pg state .
the hump structure in the its below @xmath2 is also affected by the relative height and the voltage of the superconducting coherence peak , which originates from the fact that tunneling quasiparticles in an ijj are dressed by the presence of both the sg and the pg .
since the interlayer tunneling is sensitive to the electronic state in antinodal region existing in a flat band the pg behavior in its , coexisting with the sg , should be related to the fermi surface nesting induced by the van hove singularity.@xcite
we want to acknowledge illuminating private communications with h .- y . choi .
one of us ( h .- j .
lee ) appreciates valuable discussion and communications with v. m. krasnov and a. yurgens on the thermometry in the presence of bias - induced heating .
this work was supported by the national research laboratory program administrated by korea science and engineering foundation ( kosef ) .
this paper was also supported by postech core research program and the korea research foundation grants no .
krf-2006 - 352-c0020
. 00 t. timusk and b. statt , rep .
prog . phys . *
62 * , 61 ( 1999 ) and the references cited therein .
a. damascelli , z. hussain , and z .- x .
shen , rev .
phys . * 75 * , 473 ( 2003 ) .
m. r. norman , h. ding , m. randeria , j. c. campuzano , t. yokoya , t. takeuchi , t. takahashi , t. mochiku , k. kadowaki , p. guptasarma , and d. g. hinks , nature ( london ) * 392 * , 157 ( 1998 ) ; a. kanigel , m. r. norman , m .
randeria , u. chatterjee , s. souma , a. kaminski , h. m. fretwell , s. rosenkranz , m. shi , t. sato , t. takahashi , z. z. li , h. raffy , k. kadowaki , d. hinks , l. ozyuzer , and j. c. campuzano , nature phys .
( london ) * 2 * , 447 ( 2006 ) .
d. s. dessau , z .- x .
shen , d. m. king , d. s. marshall , l. w. lombardo , p. h. dickinson , a. g. loeser , j. dicarlo , c .- h park , a. kapitulnik , and w. e. spicer , phys .
* 71 * , 2781 ( 1993 ) .
k. gofron , j. c. campuzano , a. a. abrikosov , m. lindroos , a. bansil , h. ding , d. koelling , and b. dabrowski , phys .
. lett . * 73 * , 3302 ( 1994 ) .
a. m. gabovich , a. i. voitenko , j. f. annett , and m. ausloos , supercond .
technol * 14 * , r1 ( 2001 ) . c. m. varma and l. zhu , phys .
* 98 * , 177004 ( 2007 ) ; a. p. kampf and j. r. schrieffer , phys .
b * 42 * , 7967 ( 1990 ) ; x .- g .
wen and p. a. lee , phys .
* 76 * , 503 ( 1996 ) ; i. s. elfimov , g. a. sawatzky and a. damascelli , arxiv:0706.4276v1 ; n. doiron - leyraud , c. proust , d. leboeuf , j. levallois , j .- b .
bonnemaison , r. liang , d. a. bonn , w. n. hardy and l. tailler , nature ( london ) * 447 * , 565 ( 2007 ) . m. le tacon , a. sacuto , a. geoges , g. kotliar , y. gallais , d. colson and a. forget , nat
* 2 * , 537 ( 2006 ) ; w. guyard , m. le tacon , m. cazayous , a. sacuto , a. geoges , g. kotliar , d. colson and a. forget , arxiv:0708.3731v1 ; s. h@xmath67fner , m. a. hossain , damascelli , and g. a. sawatzky , arxiv:0706.4282v1 . k. tanaka , w. s. lee , d. h. lu , a. fujimori , t. fujii , risdiana , i. terasaki , d. j. scalapino , t. p. devereaux , z. hussain , z .- x .
shen , science * 314 * , 1911 ( 2006 ) ; t. valla , a. v. fedorov , jinho lee , j. c. davis , g. d. gu , science * 314 * , 1914 ( 2006 ) .
fischer , m. kugler , i. maggio - aprile , ch .
berthod , and ch .
renner , rev .
. phys . * 79 * , 353 ( 2007 ) .
r. kleiner , f. steinmeyer , g. kunkel , and p. mller , phys .
lett . * 68 * , 2394 ( 1992 ) .
m. suzuki , t. watanabe , and a. matsuda , phys .
lett . * 82 * , 5361 ( 1999 ) ; v. m. krasnov , a. yurgens , d. winkler , p. delsing , and t. claeson , phys .
lett . * 84 * , 5860 ( 2000 ) .
a. yurgens , d. winkler , t. claeson , s. ono , and y. ando , phys .
* 90 * , 147005 ( 2003 ) ; _ ibid _ , phys .
lett . * 92 * , 259702 ( 2004 ) .
t. xiang and j. m. wheatley , phys .
lett . * 77 * , 4632 ( 1996 ) ; t. valla , a. v. fedorov , p. d. johnson , q. li , g. d. gu , and n. koshizuka , phys .
* 85 * , 828 ( 2000 ) ; m. eschrig and m. r. norman , phys .
lett . * 85 * , 3261 ( 2000 ) .
t. shibauchi , l. krusin - elbaum , ming li , m. p. maley , and p. h. kes , phys .
* 86 * , 5763 ( 2001 ) ; t. kawakami , t. shibauchi , y. terao , m. suzuki , and l. krusin - elbaum , phys .
lett . * 95 * , 017001 ( 2005 ) . v. m. krasnov , m. sandberg , and i. zogaj , phys .
* 94 * , 077003 ( 2005 ) .
x. b. zhu , y. f. wei , s. p. zhao , g. h. chen , h. f. yang , a. z. jin , and c. z. gu , phys .
b * 73 * , 224501 ( 2006 ) .
the coulomb blockade of pair tunneling may take place for a junction size below 1 @xmath16m@xmath17 , which will significantly distort the tunneling characteristics.@xcite thus , merely reducing the junction size would not be the ultimate solution of removing the self - heating .
i. latyshev , s .- j .
kim , v. n. pavlenko , t. yamashita , l. n. bulaevskii , physica c * 362 * , 156 ( 2001 ) .
bae , j .- h .
choi , and h .- j .
lee , appl .
. lett . * 86 * , 232502 ( 2005 ) .
v. m. krasnov , phys .
b * 65 * , 140504(r)(2002 ) . m .-
bae , h .- j .
lee , j. kim , and k .-
kim , appl .
. lett . * 83 * , 2187 ( 2003 ) .
t. watanabe , t. fujii , and a. matsuda , phys .
rev . lett . * 79 * , 2113 ( 1997 ) . for the same @xmath2 , an overdoped single crystal gives a smaller resistance ratio of @xmath69/@xmath70 k ) .
a limit exists in the lowest possible operation set temperature of the sample stack above the bath temperature of 4.2 k , because in our heat - compensation scheme the extra self - heating by the bias should be subtracted by reducing the heater current . with lowering the sample operation temperature
one encounters a situation where the heater current should be completely reduced , which corresponds to the lowest limit of the sample operation temperature . v.
n. zavaritsky , phys .
lett . * 92 * , 259701 ( 2004 ) ; v. n. zavaritsky , phys .
b * 72 * , 094503 ( 2005 ) .
comsol multiphysics , @xmath71comsol ab . g. k. white , proc .
a * 66 * , 559 ( 1953 ) .
m. f. crommie and a. zettl , phys .
b * 43 * , 408 ( 1991 ) . v. m. krasnov , a. yurgens , d. winkler , and p. delsing , j. appl . phys * 89 * , 5578 ( 2001 ) .
t. nakano , m. oda , c. manabe , n. momono , y. miura , and m. ido , phys .
b * 49 * , 16000 ( 1994 ) ; m. oda , k. hoya , r. kubota , c. manabe , n. momono , t. nakano , and m. ido , physica c * 281 * , 135 ( 1997 ) ; t. sato , y. naitoh , t. kamiyama , t. takahashi , t. yokoya , k. yamada , y. endoh , and k. kadowaki , physica c * 341 - 348 * , 815 ( 2000 ) .
a. mourachkine , mod .
b * 19 * , 743 ( 2005 ) .
m. tinkham , _ introduction to superconductor _ , 2nd ed .
( mcgraw - hill , new york , 1996 ) .
bae , j .- h .
choi , h .- j .
lee , and k .- s .
park , cond - mat/0512664v1 ; m .- h .
bae , j .- h .
choi , h .- j . lee and k .- s .
park , j. of kor .
48 * , 1017 ( 2006 ) .
b. w. hoogenboom , c. berthod , m. peter , @xmath68 .
fischer , and a. a. kordyuk , phys .
b * 67 * , 224502 ( 2003 ) . y. yamada and m. suzuki , phys .
b * 66 * , 132507 ( 2002 ) .
y. h. su , h. g. luo , and t. xiang , phys . rev .
b 73 , 134510 ( 2006 )
. m. hashimoto , t. yoshida , k. tanaka , a. fujimori , m. okusawa , s. wakimoto , k. yamada , t. kakeshita , h. eisaki , and s. uchida , phys .
b * 75 * , 140503(r)(2007 ) .
a. ino , c. kim , m. nakamura , t. yoshida , t. mizokawa , a. fujimori , z .- x .
shen , t. kakeshita , h. eisaki , and s. uchida , phys .
b * 65 * , 094504 ( 2002 ) ; t. yoshida , x. j. zhou , d. h. lu , seiki komiya , yoichi ando , h. eisaki , t. kakeshita , s. uchida , z. hussain , z .- x . shen , and a. fujimori , cond - mat/0610759 ( 2006 ) a. ino , t. mizokawa , k. kobayashi , a. fujimori , t. sasagawa , t. kimura , k. kishio , k. tamasaku , h. eisaki , and s. uchida , phys .
rev . lett . * 81 * , 2124 ( 1998 ) ; t. sato , t. yokoya , y. naitoh , t. takahashi , k. yamada , and y. endoh , phys .
lett . * 83 * , 2254 ( 1999 ) .
similar to its , averaging both in spatial and momentum spaces is taken for aipes .
m. suzuki and t. watanabe , phys .
* 85 * , 4787 ( 2000 ) ; s. i. vedeneev and d. k. maude , phys . rev .
b * 70 * , 184524 ( 2004 ) .
m. oda , k. hoya , r. kubota , c. manabe , n. momono , t. nakano , and m. ido , physica c * 281 * , 135 ( 1997 ) . k. ishida , k. yoshida , t. mito , y. tokunaga , y. kitaoka , k. asayama , a. nakayama , j. shimoyama , and k. kishio , phys . rev .
b * 58 * , r5960 ( 1998 ) .
t. sato , y. naitoh , t. kamiyama , t. takahashi , t. yokoya , k. yamada , y. endoh , and k. kadowaki , physica c * 341 - 348 * , 815 ( 2000 ) . v.
m. krasnov , phys .
b * 75 * , 146501 ( 2007 ) .
r. s. markiewicz , j. phys .
solids * 58 * , 1179 ( 1997 ) .
|
we report heating - compensated interlayer tunneling spectroscopy ( its ) performed on stacks of overdoped bi@xmath0sr@xmath0cacu@xmath0o@xmath1 intrinsic junctions , where most of bias - induced heating in the its was eliminated .
the onset temperature of the pseudogap ( pg ) , revealed in the hump structure of the electronic excitation spectra , reached nearly room temperature for our overdoped intrinsic junctions , which represented the genuine pg onset . at a temperature below but close to @xmath2 , both the superconducting coherence peak and the pseudogap hump coexisted , implying that the two gaps are of separate origins .
the hump voltage increased below @xmath2 , following the superconducting gap voltage , which led to a conclusion that the hump structure below @xmath2 in our its arose from the combined contribution of the quasiparticle spectral weights of two different characters ; one of the superconducting state and another of the pg state near the antinodal region .
|
the problem of distributed estimation is one of very active topics in the modern control theory and signal processing literature .
interest in this problem is motivated by a growing number of applications where a decision about the observed process must be made simultaneously by spatially distributed sensors , each taking partial measurements of the process .
when the process and measurements are subject to noise and disturbance , robustness aspects of the problem come into prominence . in the past several years
, a number of results have been presented in the literature which develop the @xmath0 control and estimation theory for distributed systems subject to uncertain perturbations ; e.g. , see @xcite . in particular , methodologies of distributed sampled - data @xmath0 filtering have been considered , e.g. , in @xcite .
that reference emphasized several distinctive aspects of realistic sensor networks , among them coupling between sensor nodes through the information communicated between neighbouring sensor nodes and the sampled nature of that coupling , which is dictated by the digital communication technology .
the latter feature of sampled data networks is an important consideration in any network design , because the amount of information that can be transmitted to / received at each node of the network is constrained , due to data rate limitations of digital communication channels . in this paper
we address some of the challenges specific to round - robin type communication protocols .
the round - robin protocol is a commonly used protocol for information transmission in networked control systems .
it allows each node to communicate with its neighbours intermittently , during scheduled time slots and is known to lead to bandwidth savings . from a hybrid systems perspective
this protocol has been studied in details in @xcite .
more recently , it has been considered in the context of time - delay systems in @xcite , where an analysis of exponential stability and @xmath1 properties of networked control systems with round - robin scheduling was presented using a delay switching system modeling . in this paper
, we further develop this technique in the context of robust distributed estimation with intermittent communications between sensing nodes .
the type of communication we consider is where the nodes broadcast their information at every scheduled time instant to all nodes in their vicinity , but they listen to only one node within their neighbourhood at a time , according to the round - robin rule . for instance , this can be achieved by encoding the transmitted information with a node - specific key .
i.e. , when node @xmath2 receives signals from multiple sources , it extracts the information sent by node @xmath3 by utilizing the key of that node , and continues doing this by rotating the keys .
the objective of this paper is to develop an algorithm for synthesis of a round - robin type protocol for a network of distributed observers , to allow this network to track dynamics of a linear uncertain plant . unlike many existing approaches to distributed estimation
, the salient feature of our methodology is that individual estimators may not be able to track the plant , if they rely solely on their own measurements , because the plant may not be observable from the node s measurements .
this issue has recently been emphasized in @xcite which demonstrated that consensus between sensors plays a crucial role in enabling individual sensor nodes to overcome lack of detectability and successfully track the plant .
necessary conditions on the network to ensure the plant is detectable / observable by the network have recently been presented in @xcite .
the first contribution of this paper is a version of the protocol of @xcite to be used with the distributed estimation schemes proposed @xcite .
we show that instead of continuously exchanging information ( the type of networks considered in those references ) , the node observers can achieve the relative @xmath0 consensus objective by exchanging information at certain sampling times , by polling one neighbour at a time .
it is assumed that sampling times are known and agreed upon at each node .
technically , this would require all nodes to have their clocks synchronized , e.g. , by means of a network time protocol @xcite .
our second contribution demonstrates that the round - robin design of @xcite can be applied to derive a network of non - switching observers .
of course , each observer periodically switches between input channels , but the observer gains remain unchanged .
this is an important feature of our methodology to ensure its scalability . in a large network of distributed estimators
, switching of observer gains typically leads to a combinatorially complex scheduling problem , and necessitates development of additional tools to resolve this complexity ; see @xcite .
we show that these issues are avoidable in observer networks of the type considered in this paper .
our main result is a sufficient condition , expressed in the form of linear matrix inequalities ( lmis ) , from which filter and interconnection gains for each node estimator can be computed , to ensure the network of sampled data observers using switching communications converges to the trajectory of the observed plant by achieving consensus between the filters at every node .
conditions for consensus of multi - agent systems using sampled data communications or systems communicating over switching graphs are well known in the literature @xcite .
this includes an observation that the sampling period has a significant effect on the system performance @xcite .
as the example presented in section [ example ] illustrates , our conditions allow to investigate the effects of intermittent sampled communications on the system performance as well . at the same time , our result provides a guarantee of the network consensus performance in the presence of disturbances .
since consensus between observers is essential for the network to be able to overcome observability / detectability limitations of individual observers @xcite , consensus performance is seen as an important design consideration which the results in this paper address . as in @xcite ,
our methodology relies on certain vector dissipativity properties of the large - scale system comprised of the observers error dynamics .
however , different form these references , to establish these vector dissipativity properties , we employ a novel class of generalized supply rates which reflect the sampled - data nature of interconnections between observers .
the general idea behind introducing such generalized supply rates can be traced to @xcite ( also , see @xcite ) , but our proposal here makes use of special properties of sampled signals . in the limit , when the maximum sampling period approaches zero , these generalized supply rate vanish , and one recovers the vector dissipativity properties of error dynamics established in @xcite .
thus , the feasibility of the conditions proposed for systems with continuously operating interconnections can be used as a preliminary ( but not conclusive ) test for the conditions proposed here .
our numerical example illustrates this point very well , showing a negligible difference between the disturbance attenuation levels obtained using the benchmark algorithm of @xcite and those obtained using our conditions under small sampling rate . at the same time
, the feasibility of our conditions makes explicit the dependence of the proposed algorithm on the sampling period .
being only sufficient conditions , our conditions are potentially conservative , but the fact that the techniques used in the derivation of these conditions showed substantial reduction of conservatism in similar problems @xcite is encouraging . the paper is organized as follows .
the problem formulation , along with the graph theory preliminaries is presented in section [ formulation ] .
the main results of the paper are given in section [ main ] .
in section [ example ] , we discuss an illustrative example , and section [ concl ] concludes the paper . [ [ notation ] ] notation + + + + + + + + throughout the paper , @xmath4 denotes a real euclidean @xmath5-dimensional vector space , with the norm @xmath6 ; here the symbol @xmath7 denotes the transpose of a matrix or a vector .
@xmath8 will denote the lebesgue space of @xmath4-valued vector - functions @xmath9 , defined on the time interval @xmath10 , with the norm @xmath11 and the inner product @xmath12 .
@xmath13 is the kronecker product of matrices , @xmath14 is the column - vector of ones .
also , @xmath15 is the determinant of @xmath16 .
consider a filter network with @xmath17 nodes and a directed graph topology @xmath18 ; @xmath19 , @xmath20 are the set of vertices and the set of edges , respectively .
the notation @xmath21 will denote the edge of the graph originating at node @xmath3 and ending at node @xmath2 . in accordance with a common convention @xcite
, we consider graphs without self - loops , i.e. , @xmath22 .
however , each node is assumed to have complete information about its filter and measurements . for each @xmath23
, we denote @xmath24 to be the _ ordered _ set of nodes supplying information to node @xmath2 , i.e , the neighbourhood of @xmath2 . without loss of generality ,
suppose the elements of @xmath25 are ordered in the ascending order .
the cardinality of @xmath25 , known as the in - degree of node @xmath2 , is denoted @xmath26 ; i.e. , @xmath26 is equal to the number of incoming edges for node @xmath2 . also , the out - degree of node @xmath2 ( .e .
, the number of outgoing edges ) is denoted @xmath27 . without loss of generality
the graph @xmath28 will be assumed to be weakly connected , that is , for every two nodes of @xmath28 there is an undirected path between these two nodes .
the rationale for this assumption is based on ( * ? ? ?
* proposition 1 ) ; according to that proposition @xmath0 consensus optimization problems over disconnected graphs are reducible to the corresponding problems over individual weakly connected components .
let @xmath29_{i , j=1}^n$ ] be the adjacency matrix of the digraph @xmath28 , i.e. , @xmath30 if @xmath31 , otherwise @xmath32 . also , let @xmath33 be the @xmath34 laplacian matrix of the graph @xmath28 , @xmath35-\mathcal{a}$ ] . in the sequel ,
a shift permutation operator defined on elements of the set @xmath25 will be used : @xmath36 furthermore , @xmath37 will denote the set obtained from @xmath25 using @xmath38 consecutive shift permutations ( [ permu ] ) . in regard to this set
, the following notation will be used throughout the paper unless stated otherwise : for @xmath39 , @xmath40 is the @xmath41-th element in the ordered set @xmath25 .
conversely , @xmath42 is the index of element @xmath3 in the permutation @xmath37 .
we will omit the superscript @xmath43 if this does not lead to ambiguity .
consider a plant described by the equation @xmath44 here @xmath45 is the state of the plant , and @xmath46 is a disturbance .
we also assume @xmath47 , so that the @xmath1-integrable solution of ( [ eq : plant ] ) with the initial condition @xmath48 exists on any finite interval @xmath49 $ ] @xcite . furthermore , it will be convenient to assume that the plant was at the state @xmath50 for all @xmath51 .
the distributed filtering problem under consideration is to estimate the state of the system ( [ eq : plant ] ) using a network of filters connected according to the graph @xmath28 .
each node takes measurements @xmath52 @xmath53 is a measurement disturbance .
as seen from ( [ u6.yi ] ) , the measurements are assumed to be taken continuously .
although in practice , measurements are usually taken at discrete time instances , we assume that the data rate of the sensors is high enough to allow for the continuous - time interpretation of the measurement signals @xmath54 .
this will enable us to focus exclusively on the effects due to sampling and intermittence of interconnections .
the measurements are processed by a network of observers connected over the graph @xmath28 .
the key assumption in this paper is to allow the observers make use of their local measurements continuously , however they can only interact with each other at discrete time instances @xmath55 , @xmath56 , with @xmath57 . for simplicity , we assume that this schedule of updates is known to all participants in the network , and therefore all nodes exchange information at the same time instance @xmath55 . however , at every time instance @xmath55 only one neighbour in the set @xmath25 is polled by each node @xmath2 , according to the ` round - robin ' rule .
formally , this leads us to define the following observer protocol : for @xmath58 , @xmath59 , @xmath60 where @xmath61 is the estimate of the plant state @xmath62 calculated at node @xmath2 , the matrices @xmath63 , @xmath64 are parameters of the filters to be determined , and @xmath65 is a given matrix .
all observers are initiated with zero initial condition , @xmath66 for all @xmath51 and all @xmath67 . in particular
, this ensures that in ( [ up7.c.d ] ) , the terms sampled at times @xmath68 are equal to zero . from now on
, we will omit the time variable when a signal is considered at time @xmath69 , and will write , for example , @xmath70 for @xmath71 .
the last term in ( [ up7.c.d ] ) reflects the desire of each node observer to update its estimate of the plant using feedback from the neighbours in its neighbourhood , according to the consensus estimation paradigm @xcite .
however , unlike these references , under the proposed protocol , only one neighbour is polled at each time @xmath55 to provide a ` neighbour feedback ' , and this sample is stored and used by the observer until time @xmath72 .
the feature of the proposed round - robin type protocol is to poll the neighbours one at a time , in a cyclic manner .
formally , this can be described by first applying the shift permutation operator @xmath73 to the neighbourhood set at every time instance @xmath55 , and then selecting the first element from the resulting permutation @xmath37 for feedback .
let @xmath74 be the local estimation error at node @xmath2 .
this error satisfies the equation : @xmath75 here we used the notation @xmath76 to represent the perturbation vector @xmath77'$ ] , and the matrices @xmath78 , @xmath79 are defined as follows @xmath80 $ ] , @xmath81 $ ] . since the plant was at the state @xmath82 for all @xmath51 , the initial conditions for ( [ e.1 ] ) are @xmath83 @xmath84 .
since the error dynamics ( [ e.1 ] ) are governed by @xmath1 integrable disturbance signals @xmath76 , we can only expect the node observers to converge in @xmath1 sense . to quantify transient consensus performance of the observer network ( [ up7.c.d ] ) under disturbances ,
consider the cost of disagreement between the observers caused by a particular vector of disturbance signals @xmath85'$ ] , @xmath86 where @xmath38 is a time - dependent index , @xmath59 , defined so that for every @xmath87 , @xmath88 .
the functional ( [ psi.l2 ] ) was originally introduced in @xcite as a measure of consensus performance of a corresponding continuous - time observer network .
it is worth noting that for each @xmath69 , @xmath89 is independent of the order in which node @xmath2 polls its neighbours , so that @xmath90 therefore , the inner summation in ( [ psi.l2 ] ) can be replaced with summation over the neighbourhood set @xmath25 .
this observation leads to the same expression for @xmath91 as in the case of continuous - time networks @xcite , @xmath92 ds.\ ] ] the following distributed estimation problem is a version of the distributed @xmath0 consensus - based estimation problem originally introduced in @xcite , modified to include the round - robin type protocol ( [ up7.c.d ] ) .
[ def1 ] the distributed estimation problem under consideration is to determine a collection of observer gains @xmath63 and interconnection coupling gains @xmath93 , for the filters ( [ up7.c.d ] ) which ensure that the following conditions are satisfied : a. in the absence of uncertainty , the interconnection of unperturbed systems ( [ e.1 ] ) must be exponentially stable .
b. the filter must ensure a specified level of transient consensus performance , as follows @xmath94 here , @xmath95 , @xmath96 is a matrix to be determined , and @xmath97 is a given constant . in @xcite , the quantity on the left - hand - side of ( [ objective.i.1 ] ) was referred to as the mean - square @xmath1 disagreement gain of the distributed observer .
note that unlike @xcite , here we aim to achieve internal stability and @xmath0 performance of the observer using a different communication protocol , which involves sampling of observer inputs according to the round - robin rule .
our approach to solving the problem in definition [ def1 ] will follow the methodology for the analysis of stability and @xmath1-gain for networked control systems proposed in @xcite . in this paper , this methodology is further extended to derive _ synthesis conditions _ for a network of observers .
the methodology in @xcite makes use of the time - delay approach to sampled - data control started in @xcite . in @xcite
the closed - loop system under consideration is presented as a switched system with multiple and ordered time - varying delays .
as can be seen from ( [ e.1 ] ) , if the observer at node @xmath2 polls a channel at time @xmath98 , the next time the same channel will be polled at time @xmath99 .
the longest time between polls of the same channel at node @xmath2 constitutes the maximum delay in communication between node @xmath2 and its neighbours , which will be denoted @xmath100 : @xmath101 the largest communication delay in the network is then @xmath102 .
it is easy to see from these definitions that @xmath103 , where @xmath104 .
consider the following lyapunov - krasovskii candidate for the system ( [ e.1 ] ) : @xmath105 where @xmath106 , @xmath107 , @xmath108 and @xmath109 , @xmath67 , are matrices and constants to be determined .
@xmath110 is a standard lyapunov - krasovskii functional used in the literature on exponential stability of systems with time - varying delays ; e.g. , see @xcite . given a matrix @xmath111 , define @xmath112 [ dissip - ineq ] suppose there exist gains @xmath64 , @xmath63 , matrices @xmath113 , and constants @xmath114 , @xmath115 , @xmath67 , such that the following vector dissipation inequality holds for all @xmath67 : for @xmath58 , @xmath116 where @xmath117 is the index of @xmath3 in the ordered permutation set @xmath37 .
then the system ( [ e.1 ] ) satisfies conditions ( i ) and ( ii ) in definition [ def1 ] .
the proof of this theorem and other statements are given in the appendix .
let @xmath118 , and define @xmath119 then , inequality ( [ vlf ] ) can be written in the standard form of a vector dissipation inequality @xcite , @xmath120 this prompts for an interpretation of @xmath121'$ ] and @xmath122'$ ] as , respectively , a vector storage function and a vector supply rate for the large scale system comprised of the error dynamics subsystems ( [ e.1 ] ) @xcite . strictly speaking , in our case such
an interpretation is somewhat artificial , since for example , the derivative signal @xmath123 is not an output of ` subsystem ' @xmath2 , and is not used for feedback by any of the neighbours this node . nonetheless , in the proof of theorem [ dissip - ineq ] the functions @xmath124 will play a role analogous to that played by generalized supply rates in @xcite .
in what follows we derive a sufficient condition for the dissipation inequality ( [ vlf ] ) to hold .
we begin with a technical lemma which essentially restates the corresponding lemma of @xcite in the form convenient for the subsequent use in the paper .
consider a vector @xmath125'$ ] , @xmath126 . also , for given @xmath127 matrices @xmath107 and @xmath128 , define @xmath129.\ ] ] [ lem1 ] suppose the matrices @xmath107 and @xmath128 are such that @xmath130\ge 0.\end{aligned}\ ] ] then @xmath131 & \ge & \delta ' \psi_i \delta.\end{aligned}\ ] ] let @xmath132'$ ] , @xmath133'$ ] , and @xmath134 be the following matrix @xmath135\otimes i.\ ] ] define @xmath136 and partition this matrix in accordance with the partition of @xmath137 : @xmath138.\ ] ] also , let us introduce the correspondingly partitioned matrix @xmath139 , \label{tildepsi}\end{aligned}\ ] ] where we let @xmath140 , @xmath141 , and @xmath142 for all other elements of @xmath143 . then the following statement holds . [ dotv.lemma ] under the conditions of lemma [ lem1 ] , @xmath144 furthermore , since @xmath145 , for every @xmath146 , we have @xmath147 this leads to the following statement .
[ pi_jv_j.lemma ] @xmath148 ' \left[\begin{array}{cc } \bar \phi_{i,11 } & \bar \phi_{i,12 } \\
\bar \phi_{i,21 } & \bar \phi_{i,22 } \end{array } \right ] \left[\begin{array}{c } \mathbf{e}_{i , t } \\
\mathbf{e}_{i , s } \end{array } \right ] , \label{pi_jv_j.1}\end{aligned}\ ] ] where @xmath149 , \\ & & \bar \phi_{i,22}= \left[\begin{array}{cccc } { \pi^2\over 4}w_{j_1 } & 0 & \ldots & 0 \\ 0 & { \pi^2\over 4}w_{j_2 } & \ldots & 0 \\ \vdots & \vdots & \ddots & \vdots \\ 0 & 0 & \ldots & { \pi^2\over 4}w_{j_{p_i } } \end{array } \right ] , \\ & & \bar \phi_{i,12}=\bar \phi_{i,21}=-\bar \phi_{i,22}.\end{aligned}\ ] ] next , we apply the descriptor method @xcite in order to derive lmis for the design of observers gains . for @xmath150 , consider the neighbourhood set @xmath25 and its corresponding permutation @xmath37 . recall that for every @xmath146 , @xmath151 is the index of node @xmath3 in the ordered set @xmath37 . according to this notation , on the interval @xmath152 the observer at node @xmath2 utilizes the sample @xmath153 , and the corresponding error equation is driven by @xmath154 .
let us define vectors @xmath155 ' , \\ & & \mathbf{e}_{i , s}=[e_{j_1}(t_{k-\nu_{j_1}^{k , i}+1})'~\ldots~ e_{j_{p_i}}(t_{k-\nu_{j_{p_i}}^{k , i}+1})']',\end{aligned}\ ] ] which consist of the current and sampled error interconnection inputs , respectively , ordered in accordance with ordering of the set @xmath25 .
note that for arbitrary compatible matrices @xmath156 , @xmath157 and @xmath158 , @xmath159 from ( [ dotvi.2 ] ) and ( [ descr ] ) it follows that @xmath160 along with condition ( [ pi_jv_j.1 ] ) established in lemma [ pi_jv_j.lemma ] , this leads to the conclusion that @xmath161 in the above inequality , @xmath162 is the vector @xmath163'$ ] , and @xmath164 is the matrix partitioned as follows @xmath165,\end{aligned}\ ] ] @xmath166 it is worth noting that the matrix @xmath164 does not depend on @xmath38 .
hence the dissipation inequality follows from the condition @xmath167 at any time @xmath69 . by combining this conclusion with theorem [ dissip - ineq ] ,
we arrive at the following statement .
[ analysis ] suppose there exist matrices @xmath106 , @xmath156 , @xmath157 , @xmath158 , @xmath168 , @xmath108 , @xmath107 , @xmath128 , constants @xmath114 , @xmath169 , and gain matrices @xmath170 , @xmath171 , which satisfy the lmi ( [ park ] ) and @xmath172 then the corresponding observer network ( [ up7.c.d ] ) solves the problem posed in definition [ def1 ] .
the matrix @xmath173 in condition ( [ objective.i.1 ] ) corresponding to this solution is @xmath174 .
theorem [ analysis ] serves as the basis for derivation of the main result of this paper , given below in theorem [ t1 ] , which is a sufficient condition for synthesis of distributed observer networks of the form ( [ up7.c.d ] ) .
consider the following matrix @xmath175,\end{aligned}\ ] ] @xmath176 [ t1 ] suppose there exists matrices
@xmath106 , @xmath156 , @xmath177 , @xmath178 , @xmath179 , @xmath108 , @xmath107 , @xmath168 , @xmath128 , and constants @xmath114 , @xmath180 , @xmath181 , @xmath182 , @xmath171 , which satisfy the lmi ( [ park ] ) and @xmath183 then the network of observers ( [ up7.c.d ] ) with @xmath184 solves the distributed estimation problem posed in definition [ def1 ] . the matrix @xmath173 in condition ( [ objective.i.1 ] ) corresponding to this solution is @xmath185 . _
proof : _ similar to @xcite , we observe that lmi ( [ lmi.analysis ] ) follows from ( [ lmi.synthesis ] ) , when we let @xmath186 , @xmath187 , and take @xmath170 to be matrices defined in ( [ gains ] ) .
then the claim of the theorem follows from theorem [ analysis ] .
@xmath188 the proposed lmi conditions involve ` free ' variables @xmath156 , @xmath157 and @xmath158 .
these variables are to reduce the conservatism of the proposed lmi conditions . at the same time they add to the number of unknowns to be used by the lmi solver . in a high - dimensional problem where this causes an excessive computational burden ,
additional constraints on these variables can be introduced to reduce the number of variables used by the solver , at the expense of a more conservative design ; e.g. , @xmath156 can be assumed to be diagonal .
consider a plant of the form ( [ eq : plant ] ) , with @xmath189 $ ] , @xmath190 $ ] .
this plant was used in the example in @xcite .
the nominal part of the plant describes one of the regimes of the so - called chua electronic circuit . to estimate this plant
, we will use the 3-node observer network shown in figure [ fig : example ] .
, when @xmath38 takes one of the values shown above the figure.,width=226 ] the measurement matrices are @xmath191 , \\ & & c_2=[-0.8986~0.1312~-1.9703 ] , \\ & & c_3=[1~0~0 ] , \quad \mbox{and}\quad d_{2i}=0 , \quad \bar d_{2i}=0.025.\end{aligned}\ ] ] with these parameters , the pairs @xmath192 and @xmath193 are not detectable , while @xmath194 is observable .
since observer at node 3 does not receive information from other observers , it acts as a conventional continuous - time @xmath0 filter , while the observers at nodes 1 and 2 utilize sampled data inputs they receive from their neighbours .
this allows them to overcome difficulties due to unstable unobservable modes of @xmath195 . for simplicity
we assume a constant sampling period of @xmath196 , so that @xmath197 . then @xmath198 , @xmath199 and @xmath200 .
we now apply theorem [ t1 ] to compute observer and interconnection gains for this system . to this end
, we solved the lmis ( [ lmi.synthesis ] ) numerically , with @xmath201 , @xmath202 , and @xmath203 ; that is , @xmath204 in this example . in fact , instead of solving the feasibility problem , we solved the optimization problem in which we sought to minimize @xmath205 subject to the lmi constraints ( [ park ] ) and ( [ lmi.synthesis ] ) .
first , we compared the performance of our method with the performance guaranteed for estimators employing continuous - time interconnections by the method in @xcite . to this end
, we set the sampling rate to a high value by letting @xmath206 . with @xmath207
, we obtained the suboptimal @xmath205 to be equal @xmath208 , which is approximately equal to the level of @xmath0 disagreement guaranteed for the comparison distributed estimator of @xcite , @xmath209 .
a slight discrepancy between the two values is likely due to numerical errors and/or conservative selection of parameters . remarkably , both algorithms assign a high gain to the @xmath0 filter at node 3 ( @xmath210'$ ] using theorem [ t1 ] versus @xmath211'$ ] using the method from @xcite ) .
next we set the sampling rate to a larger value .
after some experimenting with the tuning parameters @xmath212 , we chose @xmath213 . with @xmath214 , theorem [ t1 ]
was found to guarantee the level of @xmath0 disagreement @xmath215 , and the gain @xmath216 reduced substantially , to the value @xmath217'$ ] .
this gain is comparable with that obtained for the estimator of @xcite with this value of @xmath205 . for @xmath218 ,
the guaranteed level of @xmath0 disagreement increased substantially , to the value of @xmath219 . further increasing the sampling period to @xmath220 resulted in a prohibitively large @xmath221 .
the paper has presented a sufficient lmi condition for the design of a round - robin type interconnection protocol for networks of distributed observers .
we have shown that the proposed protocol allows one to use sampled - data communications between the observers in the network , and does not require a combinatorial gain scheduling . as a result ,
the node observers are shown to be capable of achieving the @xmath0 consensus objective introduced in @xcite .
as our example demonstrates , the proposed round - robin protocol achieves this objective at the expense of moderately deteriorated performance .
define the vector function @xmath222'$ ] and the matrix @xmath223 $ ] , @xmath224 , @xmath225 if @xmath226 , and @xmath227 if @xmath228 and @xmath229 .
it follows from ( [ vlf ] ) that for @xmath58 , @xmath230 by changing the order of summation in the second term , we further obtain @xmath231 hence , on the time interval @xmath58 , @xmath232\bigg ] .
\label{vlf.sum.1}\end{aligned}\ ] ] let @xmath233 be a time instant , @xmath234 .
let us fix @xmath2 and @xmath146 , and consider the partition of the interval @xmath49 $ ] into subintervals @xmath235&=&[0,t_{\bar k-\bar dp_i-\bar\nu+1})\cup [ t_{\bar k-\bar\nu+1},t ) \nonumber \\ & & \cup\left(\bigcup_{d=1}^{\bar d } [ t_{\bar k - dp_i-\bar\nu+1},t_{\bar k-(d-1)p_i-\bar\nu+1})\right ) , \label{partition}\end{aligned}\ ] ] where @xmath236 is the index of @xmath3 in the permutation @xmath237 , and @xmath238 is the largest integer number such that @xmath239 .
note that @xmath240 , and @xmath241 .
the significance of this partition is that on each interval @xmath242 , the observer at node @xmath2 makes use of the sample @xmath243 .
therefore the input @xmath244 into the error dynamics equation ( [ e.1 ] ) at node @xmath2 holds the constant value @xmath245 over this interval of time , with @xmath246 for @xmath247 .
that is , for @xmath248 , @xmath249 where @xmath250 is determined from the condition @xmath88 , and @xmath251 is determined accordingly , as an index of @xmath3 in the permutation @xmath252 . using the wirtinger s inequality @xcite
, it follows that @xmath254 similarly , @xmath255 therefore , we conclude from ( [ sum ] ) , ( [ i m a ] ) and ( [ ima.0 ] ) that @xmath256 hence , it follows from ( [ vlf.sum.1 ] ) that @xmath257 the statement of the theorem then follows from ( [ vlf.sum.2 ] ) using the same argument as that used in the proof of theorem 1 in @xcite . consider @xmath258 : @xmath259ds \\ & & -2\alpha_i\int_{t-\tau_i}^t e^{-2\alpha_i(t - s ) } e_i(s)'s_ie_i(s)ds \\ & & - 2\alpha_i \tau_i \int_{t-\tau_i } ^t e^{-2\alpha_i(t - s ) } \dot e_i(s)'(\tau + s - t)r_i \dot e_i(s)ds.\end{aligned}\ ] ] t. hayakawa , t. matsuzawa , and s. hara .
formation control of multi - agent systems with sampled information relationship between information exchange structure and control performance . in _ proc .
45th ieee conference on decision and control _ , pages 43334338 , 2006 .
w. p. m. h. heemels , a. r. teel , n. van de wouw , and d. nesic .
networked control systems with communication constraints : tradeoffs between transmission intervals , delays and performance . , 55(8):17811796 , 2010 .
|
the paper considers a distributed robust estimation problem over a network with directed topology involving continuous time observers .
while measurements are available to the observers continuously , the nodes interact according to a round - robin rule , at discrete time instances .
the results of the paper are sufficient conditions which guarantee a suboptimal @xmath0 level of consensus between observers with sampled interconnections . large - scale systems , distributed estimation , robust observers , consensus , vector dissipativity , sampled - input systems , time - delay systems .
|
several texts provide an elementary derivation of the kinematics of special relativity , eventually based on the first part of einstein s fundamental paper of 1905 .
@xcite starting from the two postulates of the total equivalence of inertial reference systems , and of the constancy of the speed of light in all such systems , it is in fact easy to obtain the expression of the lorentz transformation of space - time coordinates .
a particularly simple and appealing derivation is obtained by exploiting bondi s so - called @xmath0-calculus , @xcite itself based on the doppler effect .
nevertheless going from kinematics ( the lorentz transformation ) to dynamics , and in particular to the relativistic expressions of momentum and energy , is often achieved by resorting to more sophisticated concepts , like that of a four - vector , or by the use of quantum considerations .
( one example of this approach is the `` elementary derivation '' suggested by f. rohrlich , @xcite which exploits the expressions of the momentum and energy of a photon of frequency @xmath1 . )
einstein himself had originally derived the mass - energy equivalence via an explicit use of electrodynamics , @xcite and not just by the kind of kinematic considerations which he had developed in the first part of his 1905 paper , which are based on the constancy of the speed of light , but do not otherwise depend on maxwell s equations .
einstein had remarked this problem , and he proposed in 1935 an elementary derivation of the mass - energy relation , independent of his 1905 argument , motivating it with the following words : @xcite _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the special theory of relativity grew out of the maxwell electromagnetic equations .
so it came about that even in the derivation of the mechanical concepts and their relations the consideration of those of the electromagnetic field has played an essential role .
the question as to the independence of those relations is a natural one because the lorentz transformation , the real basis of the special relativity theory , in itself has nothing to do with the maxwell theory and because we do not know the extent to which the energy concepts of the maxwell theory can be maintained in the face of the data of molecular physics . in the following considerations , except for the lorentz transformation , we will depend only on the assumption of the conservation principles for impulse and energy .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ einstein s considerations exploit a conceptual experiment introduced by g.n . lewis and r.c .
tolman @xcite and further discussed by p.s .
epstein , @xcite where one considers collisions between pairs of particles in different inertial reference frames , and one looks for the expressions of momentum and energy by postulating their conservation .
it is interesting to point out that lewis and tolman s paper , as well as epstein s one , only consider elastic collision and derive the relativistic expression of momentum , while they provide a doubtful argument for the mass - energy equivalence by the consideration of the change of the `` relativistic mass '' with speed .
`` relativistic mass '' is in fact a rather problematic concept,@xcite and there is a growing consensus to avoid its introduction in the teaching of relativity .
einstein derives instead the equivalence by simply extending the argument to inelastic collisions .
the advantage of this approach for introducing the basic concepts of special relativity has been well remarked by r.f .
feynman who , in his _ lectures _ ( ref . ) , derives the relativistic expressions of momentum and energy in a way that closely resembles einstein s one .
( feynman s derivation is however marred by his use of the `` relativistic mass '' . )
einstein s argument has been more recently discussed by f. flores , @xcite who identifies three closely related but different claims within the mass - energy equivalence concept , and compares einstein s 1935 argument with his original 1905 derivation @xcite and with m. friedman s 1983 derivation , @xcite which rests upon the consideration of newton s equations in special relativity . in this note ,
i present this line of thought in the hope that it may be found useful for the presentation of these fundamental concepts of special relativity in introductory courses for students of physics and mathematics . while the derivation of the relativistic expression of momentum in sec .
[ sec : qdm ] is close to epstein s and feynman s arguments , the discussion of the expression of the kinetic energy and of the mass - energy equivalence is closer to einstein s one .
following einstein s 1905 paper ( ref . ) , the kinematic concepts of special relativity rest on the following postulates : 1 .
the laws that govern the transformations of the state of physical systems take the same form in reference frames animated by uniform translational motion one with respect to the other .
2 . there is a class of such reference frames ( the _ inertial frames _ ) in which the speed of light assumes the same value @xmath2 , independently of the state of motion of its source .
we shall choose from now on units in which @xmath3 and limit our considerations to inertial frames .
based on these postulates one can easily derive lorentz transformations in the following form . let us consider two reference frames , @xmath4 and @xmath5 , such that @xmath5 is in uniform translational motion in the positive @xmath6 direction and with speed @xmath7 with respect to @xmath4 . then the event of coordinates @xmath8 in @xmath5 has in @xmath4 the coordinates @xmath9 , where @xmath10 and we have defined @xmath11 the same relation holds for the differentials @xmath12 , @xmath13 , etc . dividing by @xmath12 we obtain the rules for the transformation of velocities : @xmath14 to introduce dynamical concepts we obviously need supplementary postulates .
we shall therefore postulate the following : 1 .
the momentum @xmath15 and the energy @xmath16 of a particle possessing the velocity @xmath17 in the reference frame @xmath4 have respectively the expressions @xmath18 where @xmath19 is a constant , that can be called the _
rest energy _ , @xmath20 is a positive constant ( which is _ relativistically invariant _ , i.e. , does not change from one inertial frame to another , and thus _ does not depend on the particle s velocity _ ) which we shall simply call its _ mass _ , and @xmath21 and @xmath22 are _ monotonically increasing _ universal functions of @xmath23 .
the fact that the functions @xmath24 and @xmath25 depend only on the magnitude @xmath26 of the velocity , but not on its direction , can be inferred by the isotropy of space .
2 . for @xmath27
these expressions reduce to the well - known classical ones .
one has in particular @xmath28 3 .
the total momentum @xmath29 and the total energy @xmath30 of a system of several particles are respectively given by the sum of @xmath15 and of @xmath16 running over all the particles of the system .
* conservation of momentum and energy : * let us assume that ( elastic or inelastic ) collisions occur in a system of particles . then , in each reference frame , @xmath29 and @xmath30 maintain the same values before and after each collision .
as a corollary , the velocity of each particle remains constant as long as no collision occurs , because one can consider systems made up of single , independent particles .
let us now consider a _ particle pair _
, i.e. , a system made of two particles with equal values of @xmath20 . let us assume that in a reference frame @xmath4 they have opposite velocities @xmath31 , @xmath32 , where @xmath33 , with @xmath34 , @xmath35 .
thus @xmath36 .
let us moreover assume that the particles undergo an elastic collision , and take on respectively the velocities @xmath37 and @xmath38 after it . by the conservation of momentum one must have @xmath39 and @xmath40 , _ independently of the form of the function @xmath21 .
_ indeed , one has @xmath41 before the collision . by the conservation law
one must have @xmath42 thus the vectors @xmath43 and @xmath44 are antiparallel , and one has @xmath45 since the function @xmath21 is monotonically increasing , this equation can only be satisfied if @xmath46 , and we have therefore @xmath47 .
now energy conservation imposes @xmath48 .
indeed , the total kinetic energy before the collision is given by @xmath49 , and after it is given by @xmath50 .
since @xmath22 is a monotonically increasing function of @xmath26 , this condition can only be satisfied if @xmath51 .
let us now consider the special case where the velocity change is parallel to the @xmath52 axis .
( this rules out `` head - on collisions '' , where the particles simply exchange their velocity . )
let the particles velocities be given by @xmath53 and @xmath54 ( cf .
[ fig : coll ] ) in the @xmath5 reference frame .
let us now look at the same collision in a reference frame @xmath4 moving with a velocity @xmath55 in the @xmath6 direction with respect to @xmath5 . reference frame .
below : the same collision as seen in the @xmath4 reference frame .
figures are schematic and no quantitative relations are implied.,title="fig:",scaledwidth=45.0% ] + reference frame .
below : the same collision as seen in the @xmath4 reference frame .
figures are schematic and no quantitative relations are implied.,title="fig:",scaledwidth=45.0% ] in this reference frame , the velocities @xmath56 of the particles are respectively given by @xmath57 where @xmath58 one can easily check that @xmath59 and therefore that @xmath60 one can also obtain this result by applying eq .
( [ eq : velocities ] ) to the transformation from the @xmath5 reference frame to a frame @xmath61 animated , with respect to @xmath5 , by a uniform translational motion with velocity @xmath62 parallel to the @xmath6 axis . in this frame
the @xmath6 component of the velocity @xmath63 vanishes and that of the @xmath64 velocity is equal to @xmath7 , and thus the speeds of the two particles are interchanged .
let us now consider the conservation of momentum .
the change @xmath65 of the momentum of particle 1 is given by @xmath66 where @xmath67 is the @xmath52 axis versor .
the corresponding quantity for particle 2 is given by @xmath68 where @xmath69 .
let us momentarily assume @xmath70 : then @xmath71 and @xmath72 . from momentum conservation
we obtain @xmath73 , which implies @xmath74 since @xmath75 , we obtain @xmath76 having obtained this result for @xmath70 , it is easy to see that it also holds for larger values of @xmath77 and @xmath78 , by substituting @xmath7 with the speed of the corresponding particle .
we have in fact @xmath79 from which it follows @xmath80 namely @xmath81 an identity which is easy to check directly .
let us consider a particle having the velocity @xmath82 , parallel to the @xmath6 axis , in the @xmath5 reference frame .
its velocity @xmath17 in the @xmath4 frame , in uniform translational motion with respect to @xmath5 with speed @xmath7 in the direction of the positive @xmath6-axis , is given by @xmath83 , with @xmath84 one can easily see that @xmath85 if @xmath86 is not parallel to the @xmath6 axis , but one has instead @xmath87 , one has the more general relation @xmath88 which can be obtained with a little algebra . it is also easy to check that @xmath89 let us now consider a particle pair , which have opposite velocities @xmath90 , @xmath54 in the @xmath5 frame .
thus @xmath91 is the common value of the particles speed in the @xmath5 frame .
let us denote by @xmath31 and @xmath92 the corresponding velocities in the @xmath4 frame .
we obtain @xmath93 we have seen that an elastic collision in the @xmath4 frame can not change the common value @xmath94 of the particles speed .
thus the right - hand side of this equation can not change in the collision .
but then neither can its left - hand side .
if we denote by @xmath43 and @xmath44 the particles speeds in the @xmath4 frame after the collision , we obtain @xmath95 as einstein ( ref . )
points out , these equations have the form of conservation laws .
we can thus interpret@xcite @xmath96 as the kinetic energy @xmath97 of a particle with mass @xmath20 animated by a velocity @xmath17 . for small values of
@xmath26 this quantity is thus given by @xmath98 in agreement with the classical limit .
we can thus set @xmath99 let us remark moreover that , by applying eqs .
( [ eq : momentum ] ) to a particle pair , we obtain @xmath100 we can thus derive the following relation : @xmath101 which can be interpreted as the conservation law for the momentum . we thus recover the relativistic expression of the momentum derived in sec .
[ sec : qdm ] .
let us now consider a totally inelastic collision in a particle pair . in the reference frame @xmath5 in which @xmath102 , the total kinetic energy before the collision is given by @xmath103 for simplicity , we shall assume that the particles velocities @xmath104 are parallel to the @xmath52 axis
. the total kinetic energy after the collision vanishes , but the energy of the resulting particle has increased by @xmath105 . by momentum conservation
the resulting particle is at rest in the reference frame @xmath5 and has thus velocity @xmath106 parallel to the @xmath6 axis in the @xmath4 frame .
let us denote by @xmath107 its mass .
in the @xmath4 frame the total momentum before the collision is given by @xmath108 while after the collision it has the value @xmath109 we obtain therefore @xmath110 therefore when two particles collide inelastically to form a new particle , the mass of the resulting particle is larger than the sum of the masses of the colliding particles by a quantity exactly equal , in our units , to the kinetic energy transformed by the collision into other energy forms .
one can thus introduce a `` natural '' choice of the energy at rest @xmath19 of a particle , by equating it ( in our units ) to its mass , also because , `` from the nature of the concept , [ that ] is determined only to within an additive constant , one can stipulate that @xmath19 should vanish together with @xmath20.''@xcite .
we can then identify @xmath111 with the total energy of a particle , and associate the change @xmath112 of energy from the kinetic to a different form with a change @xmath113 of the mass of the particle . in the more general case , in which the two particles do not coalesce , but are animated by velocities with the same magnitude @xmath114 in the @xmath5 frame after the collision , one can then express the conservation of energy by @xmath115 where @xmath116 is the mass of each particle after the collision .
the same applies in the @xmath4 frame , if we take into consideration the relation @xmath117 which holds since both @xmath31 and @xmath92 are parallel to the @xmath52 axis .
it is then a simple matter to verify that the energy @xmath16 and the momentum @xmath15 of a particle of mass @xmath20 satisfy the relation @xmath118 where the right - hand side is relativistically invariant .
this relation also holds in the limit of a zero - mass particle , for which it assumes the form @xmath119 as a corollary , _ mass is not additive _ :
the mass of a system of particles depends on its total energy content in the vanishing - momentum frame .
thus when light waves ( of vanishing mass ) with opposite momenta are emitted from a particle at rest , the mass of the particle changes ( as argued in einstein s 1905 work , ref . ) . since the right - hand side of eq .
( [ eq : epm ] ) is invariant , it can be taken as the starting point to show that the energy @xmath16 and the momentum @xmath15 combine into a relativistic four - vector , a property which generalizes to particle systems .
one should point out that this line of thought does not rest on maxwell s equations ( keeping only the constancy of the speed of light ) and neither does on other mechanical concepts , in particular on the concept of force , that is difficult to justify in special relativity .
einstein criticizes the use of the concept of force in the derivation of the relativistic expression of momentum , contained in the book by g.d .
birkhoff and r.e .
langer , _ relativity and modern physics _ , @xcite exactly for this reason , as made clear by the closing paragraphs of his 1935 paper : @xcite _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ thus , in the book just mentioned , essential use is made of the concept of force , which in the relativity theory has no such direct significance as it has in classical mechanics .
this is connected with the fact that , in the latter , the force is to be considered as a given function of the coordinates of all the particles , which is obviously not possible in the relativity theory .
therefore i have avoided introducing the force concept .
furthermore , i was concerned with avoiding making any assumption concerning the transformation character of impulse and energy with respect to a lorentz transformation .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 99 a. einstein , `` zur elektrodynamik bewegter krper '' , annalen der physik * 17 * 891921 ( 1905 ) .
h. bondi , _ relativity and common sense _
( dover ed . )
( dover , new york , 1980 ) .
f. rohrlich , `` an elementary derivation of @xmath120 '' , am . j. phys . * 58 * 348349 ( 1990 ) .
a. einstein , `` ist die trgheit eines krpers von seinem energieinhalt abhngig ? '' , annalen der physik * 18 * 639641 ( 1905 ) .
a. einstein , `` elementary derivation of the equivalence of mass and energy '' , bull .
41 * 223230 ( 1935 ) .
g. n. lewis and r. c. tolman , `` the principle of relativity and non - newtonian mechanics '' , phil . mag . * 6 * 510523 ( 1909 ) . p. s. epstein , `` ber relativistische statik '' , annalen der physik * 36 * 779795 ( 1911 ) . c. g. adler , ``
does mass really depend on velocity , dad ? ''
am . j. phys .
* 55*(8 ) 739743 ( 1987 ) . l. b. okun , `` the concept of mass '' , physics today * 42*(6 ) 3136 ( 1989 ) ; `` the mass versus relativistic and rest masses '' , am .
j. phys . * 77*(5 ) 430431 ( 2009 ) .
r. feynman , r. leighton and m. sands , _ the feynman lectures on physics _ ( addison - wesley , boston , 1964 , 1966 ) . f. flores , `` einstein s 1935 derivation of @xmath120 '' , stud .
phys . * 29*(2 ) 223243 ( 1998 ) .
m. friedman , _ foundations of spacetime theories _ ( princeton university press , princeton , 1983 ) .
equation ( [ eq : consgamma ] ) implies the conservation of @xmath121 .
the value of the constant is chosen so that the expression vanishes for @xmath122 .
g. d. birkhoff and r. e. langer , _ relativity and modern physics _
( harvard university press , cambridge ( ma ) , 1923 ) .
|
the derivation of the expressions of momentum and energy of a particle in special relativity is often less than satisfactory in elementary texts . in some ,
it is obtained by resorting to quantum or electrodynamic considerations , in others by introducing less - than - elementary concepts , like that of a four - vector , or even misleading ones , like `` relativistic mass '' .
nevertheless it is possible , following ideas described by einstein in 1935 , to obtain a fully elementary derivation of these expressions based only on the lorentz transformations , on the conservation laws , and on the newtonian limit .
the resulting argument allows for a clearer and logically consistent introduction to the basic concepts of relativistic dynamics .
|
certain anisotropic molecules such as n@xmath2 , c@xmath3 , and kcn form a cubic crystal and , at lower temperatures , they undergo an orientation phase transition with a specific - heat peak @xcite , where the crystal structure changes to a noncubic one . furthermore , mixtures of anisotropic particles @xcite such as ( kcn)@xmath4(kbr)@xmath5 and one - component systems of globular molecules @xcite such as ethanol and cyclohexanol become orientation glass . in such glass , the phase ordering should occur only on small spatial scales with mesoscopically heterogeneous orientation fluctuations .
because of anisotropic molecular shapes , there should be a direct ( proper ) coupling between the molecular orientations and the lattice deformations @xcite .
in fact , the shear modulus becomes small around the orientational order - disorder or glass transition @xcite .
these systems thus exhibit singular acoustic and plastic behaviors @xcite , but there has been no systematic experiment in the nonlinear response regime .
many of these anisotropic molecules have dipolar moments also , yielding dielectric anomaly near the transition . in metallic alloys
, a structural phase transition arises from the displacements of the atoms in each unit cell from their equilibrium positions in the high - symmetry phase .
some alloys undergo a martensitic phase transition gradually from a high - temperature phase to a low - temperature phase over a rather wide temperature range and , at sufficiently low temperatures , they are composed of multiple martensitic variants or domains @xcite .
in particular , a system of off - stoichiometric intermetallic ti@xmath6ni@xmath7 has been studied extensively @xcite .
even at @xmath8 , it becomes strain glass , exhibiting the shape - memory effect and the superelasticity , where strain heterogeneities with sizes of order 10 nm were observed @xcite . as a similar example ,
metallic ferroelectric glass , called relaxor , exhibits large dielectric response to applied electric field @xcite , where the electric polarization and the lattice deformations are coupled and frozen polar nanodomains are produced in the presence of the compositional disorder in the perovskite structure . in soft matter , impurities often strongly disturb or influence phase transitions .
examples of impurities are filler particles in phase - separating polymer blends @xcite , microemulsions in nematic liquid crystals @xcite , and crosslink irregularities in polymer gels @xcite . for gels , some authors developed random crosslink models @xcite .
moreover , in gels with liquid crystal solvents @xcite , the isotropic - nematic phase transition is analogous to the orientation phase transition in solids , where the coupling between the molecular orientation and the elasticity leads to singular elastic behavior . in such liquid crystal gels ,
nematic polydomains are produced by random crosslinkage and polydomain - monodomain transitions are induced by applied stress or electric field @xcite , as numerically studied by uchida @xcite using quenched random stress .
the polydomains obviously correspond to the mesoscopic orientation heterogeneities in solids .
we also mention experiments of crystal formation and glass transition using elongated colloidal particles in three dimensions @xcite and in two dimension @xcite .
the mesoscopic heterogeneities produced by impurities are widely recognized in various solid and soft materials .
we may mention two previous approaches .
one is based on a random field coupled to the order parameter @xcite . in particular , vasseur and
lookman @xcite introduced a spin glass theory supplemented with the elastic interaction ( the long - range interaction among the order parameter @xmath9 mediated by the elastic deformations ) @xcite .
the other is a phase - field ( ginzburg - landau ) theory with a random critical temperature and the elastic interaction @xcite , where the quadratic term ( @xmath10 ) in the free energy has a random coefficient . in these theories ,
the impurities are governed by an artificial random distribution without spatial correlations .
therefore , they lack microscopic physical pictures of the impurity disordering . from our viewpoint ,
microscopic approaches are particularly needed when each impurity strongly perturbs the local order parameter . to perform first - principle calculations of the mesoscopic heterogeneities , we start with lennard - jones systems composed of anisotropic host particles and impurities to create orientationally disordered and ordered crystal states .
( i ) in such states we may examine the degree of heterogeneities by changing the impurity composition .
( ii ) we may describe a tendency of impurity clustering or aggregation @xcite , which depends on the cooling rate from liquid .
it apparently governs the degree of vitrification , for example , in water containing a considerable amount of salt @xcite .
( iii ) we also note that the positional disorder and the orientation disorder have been discussed separately in the literature . in this paper , they appear simultaneously , though the former is weaker than the latter .
the organization of this paper is as follows . in sec.ii
, we will present the backgrounds of our theory and simulation . in sec.iii
, we will give simulation results for one - component systems of elliptic particles forming crystal to examine the orientation transition .
in sec.iv , we will treat mixtures of elliptic particles and larger repulsive impurities , where the elliptic particles are aligned in the planar alignment around the impurities@xcite . in sec.v
, we will examine the orientation dynamics of the elliptic particles .
in sec.vi , we will treat small attractive impurities , which tend to form aggregates and solvate several elliptic particles in the homeotropic alignment@xcite .
we propose a simple microscopic model of binary mixtures in two dimensions , which exhibits orientation phase transitions and glass behavior .
we do not introduce the dipolar interaction supposing nonpolar molecules . in our model , the first and second components are composed of elliptic and spherical particles , respectively .
their numbers are @xmath11 and @xmath12 , where @xmath13 in this paper .
the composition is defined by c = n_2/n , which is either of 0 , 0.05 , 0.1 , 0.15 , 0.2 , or 0.3 in this paper .
thus the particles of the second component constitute impurities .
the particle positions are written as @xmath14 ( @xmath15 ) .
the orientation vectors of the elliptic particles may be expressed in terms of angles @xmath16 as _
i=(_i,_i ) , where @xmath17 . the pair potential @xmath18 between particles @xmath19 and @xmath20 ( @xmath21 ) is a truncated modified lennard - jones potential .
that is , for @xmath22 it is zero , while for @xmath23 it reads u_ij = 4-c_ij . here , @xmath24 with @xmath25 . in terms of the diameters @xmath26 and @xmath27 of the two species , we define _
= ( _ + _ ) /2 . in eq(3 ) , @xmath28 is the value of the first term at @xmath29 , ensuring the continuity of @xmath18 .
the @xmath30 is the characteristic interaction energy .
the particle anisotropy is taken into account by the angle factors @xmath31 and @xmath32 , which depend on the relative direction @xmath33 and the orientations @xmath34 and @xmath35 of the elliptic particles .
there can be a variety of their forms depending on the nature of the anisotropic interactions . throughout this paper ,
we assume the following form , a_ij = _ 1 ( n_i)^2 + _ 1 ( n_j)^2 , where @xmath36 is the anisotropy strength of repulsion . the @xmath37 ( @xmath38 ) is equal to 1 for @xmath39 @xmath40 ) and 0 for @xmath41 ( @xmath42 ) .
thus , in the right hand side , the first ( second ) term is nonvanishing only when @xmath43 ( @xmath44 ) belongs to the first species . in sec.iv
, we treat large spherical impurities repelling the elliptic particles by setting @xmath45 and @xmath46 .
if @xmath47 in this case , there appears a tendency of parallel alignment of the elliptic particles at the impurity surfaces @xcite . on the other hand , in sec.vi , we assume @xmath48 and b_ij = _ 1_2 ( n_i)^2 + _ 2_1 ( n_j)^2 , where @xmath49 is the anisotropy strength of attraction . in this case , the attractive interaction is anisotropic only between the elliptic particles and small spherical impurities and , if @xmath50 , there appears a tendency of homeotropic alignment @xcite at the impurity surfaces .
the total energy is written as @xmath51 , where @xmath52 is the potential energy and @xmath53 is the kinetic energy , u & = & _ 1i < jnu_ij , + k & = & _ 1 in |_i|^2 + _
1in_1 |_i|^2 , where @xmath54 , @xmath55 , @xmath56 and @xmath57 are the masses , and @xmath58 is the moment of inertia of the first component . in this paper
, we set @xmath59 .
the newton equations of motion are now written as & & m _ _ i = k =- u , + & & i_1 _ i = k= -u , where @xmath60 , @xmath61 , and @xmath62 .
the second line holds for the first component ( @xmath63 ) .
however , since we treat equilibrium or at least nearly steady states , we attach a nos@xmath64-hoover thermostat @xcite to all the particles by adding the thermostat terms in eqs.(9 ) and ( 10 ) .
unless confusion may occur , space , time , and temperature will be measured in units of @xmath65 , _ 0 = _ 1 , and @xmath66 , respectively , where @xmath67 is the boltzmann constant .
stress ( and elastic moduli ) will be measured in units of @xmath68 . from eqs.(3 ) , ( 5 ) , and ( 6 ) the elliptic particles have angle - dependent diameters .
let the particles @xmath43 and @xmath44 belong to the first species .
then minimization of @xmath18 in eq.(3 ) with respect to @xmath69 gives @xmath70 .
thus the shortest diameter @xmath71 is given for the perpendicular orientations ( @xmath72 ) , while the longest diameter @xmath73 by the parallel orientations ( @xmath74 ) so that a_s= 2 ^ 1/6_1,a_= ( 1 + 2)^1/62 ^ 1/6_1 .
the ratio of these lengths ( the aspect ratio ) is given by @xmath75 .
for example , @xmath76 is equal to @xmath77 for @xmath78 and to 1.23 for @xmath79 .
we estimate the effective molecular area and the momentum of inertia of the elliptic particles as s_1= a_s a_/4 , i_1= ( a_^2 + a_s^2 ) m_1 /4 . in this paper
, we fix the overall packing fraction as _
pack=(n_1s_1 + n_2 s_2)/v=0.95 , where @xmath80 and @xmath81 is the system volume
. then the system length is @xmath82 .
our potential is analogous to the gay - berne potential for anisotropic molecules used to simulate mesophases of liquid crystals @xcite and the shintani - tanaka potential with five - fold symmetry used to study frustrated particle configurations at high densities@xcite .
it is worth noting that angle - dependent potentials have been used for lipids forming membranes .
@xcite . for each particle @xmath43 of the first species ( @xmath83 )
, we may introduce the orientation tensor @xmath84 ( @xmath85 ) in terms of the orientation vectors @xmath86 as _ i&= & ( 1+n_b^i)^-1 ( n_in_i + _ j n_j n_j ) -/2 + & = & q_i ( d_i d_i -/2 ) , where @xmath87 is the unit tensor and @xmath88 is the director with @xmath89 .
the summation is over the bonded particles @xmath90 ) of the first species with @xmath91 being the number of these particles of order 20 .
if a hexagonal lattice is formed , it includes the second nearest neighbor particles .
the angle of the director is defined by _
i=(_i,_i ) , in the range @xmath92 .
the amplitude @xmath93 is given by q_i^2= 2 _ , q_i^2 .
we will calculate the average over the elliptic particles , = _
1i n_1 q_i^2/n_1 , which represents the overall degree of orientation order .
the angle @xmath94 varies more smoothly than @xmath16 , but they coincide in ordered domains at low @xmath0 .
the @xmath95 is of order 0.1 in disordered states due to the thermal fluctuations , but it increases up to unity within domains at low @xmath0 . as a merit in visualization , @xmath95 is small in the interface regions at low @xmath0 ( see the right panels of fig.1 ) .
since the tensor @xmath96 is symmetric and traceless , its components are written as @xmath97 and @xmath98 . in terms of @xmath94
we have q_i2=q_i ( 2_i ) , q_i3=q_i ( 2_i ) .
these variables change with respect to a rotation of the reference frame by an angle @xmath9 as @xcite q_i2&=&q_i2(2 ) + q_i3(2 ) , + q_i3&=&q_i3(2 ) -q_i2(2 ) .
we also introduce the following density variables as q_2(r ) & = & _ i1 q_i2 ( r_i - r ) , + q_3(r ) & = & _ i1 q_i3 ( r_i - r ) .
we will calculate the following structure factor , s_q(k)&= & + & = & , where @xmath99 and @xmath100 are the fourier components of @xmath101 and @xmath102 , respectively . from eq.(20 ) the structure factor of @xmath103 and that of @xmath104 coincide under the rotational invariance of the system ( without stretching ) , leading to the second line of eq.(23 ) .
if there is no anisotropic overall strain , the isotropy holds for @xmath105 much larger than the inverse system length , leading to eq.(23 ) and @xmath106 . in the range @xmath107 ( left ) and order parameter amplitudes @xmath108 ( right ) of all the particles on a lattice in the @xmath109 plane with @xmath110 and @xmath78 for @xmath111 , @xmath112 , @xmath113 , and @xmath114 from above .
as @xmath0 is lowered , orientation order develops gradually with lattice deformations . ] around a junction point of three variants in the bottom panel of fig.1 at @xmath115 .
the angles among the three lines are @xmath116 , @xmath117 , and @xmath118 degrees , being approximately multiples of @xmath119 .
right : hexagonal lattice structure in an ordered variant composed of isosceles triangles for @xmath78 and @xmath115 . ] in eq.(23 ) vs @xmath105 with @xmath110 and @xmath78 for @xmath120 , and @xmath121 , which grows algebraically for small @xmath122 in the bkt phase .
right : average amplitude @xmath123 in eq.(18 ) vs @xmath0 for @xmath124 , @xmath125 , @xmath126 , and @xmath127 , which increases gradually but steeply at low @xmath0 . ]
in this section , we treat pure ( one - component ) systems of the elliptic particles ( @xmath110 ) .
we assume not large values of @xmath128 such that the crystallization first occurs at @xmath129 with random molecular orientations . far below @xmath130
, we study an orientation phase transition on a hexagonal lattice and singular mechanical behavior specific to multi - variant states .
a number of authors @xcite numerically examined the phase behavior of one - component hard rod systems in three dimensions in the plane of the aspect ratio and the density .
if the particles are rather close to spheres , they found orientationally disordered and ordered crystal phases .
solids in the orientationally disordered phase have been called `` plastic solids '' @xcite . to understand singular mechanical properties of tini around its martensitic phase transition , ding
_ et al . _
@xcite performed molecular dynamics simulation on mixtures of two species of spherical particles . in figs.1 - 3 , we show our simulation results at fixed volume under the periodic boundary condition . assuming a nos@xmath64-hoover thermostat @xcite
, we started with a liquid at @xmath131 , quenched the system to @xmath132 below the melting , and annealed it for @xmath133 .
we then lowered @xmath0 to a final low temperature . here , even if the cooling rate was varied after the crystal formation ( in the range @xmath134 ) , essentially the same results followed .
that is , there was no history - dependent behavior . in fig.1
, we show the orientation angles @xmath16 of all the particles in the range @xmath107 ( left ) and the order parameter amplitudes @xmath108 in eq.(17 ) ( right ) at @xmath135 , @xmath113 , and @xmath114 . with lowering @xmath0
, three equivalent variants emerge due to the underlying hexagonal lattice .
their areal fractions are all nearly equal to @xmath136 .
for @xmath137 the time scale of the patterns is of order @xmath138 , while for @xmath139 and @xmath114 the patterns are frozen even on time scale of @xmath140 . at low @xmath0 , the junction angles , at which two or more domain boundaries intersect , are multiples of @xmath119 .
as illustrated in fig.2 , this geometrical constraint arises from the orientation - lattice coupling .
it serves to pin the domain growth at a characteristic size even without impurities @xcite .
similar pinned domain patterns have been observed on hexagonal planes @xcite and were reproduced by phase - field simulation @xcite .
we may define the surface tension @xmath141 on the interfaces far from the junction regions . in our model ,
@xmath142 for @xmath78 and @xmath143 for @xmath79 at low @xmath0 . in fig.3 ,
the structure factor @xmath144 in eq.(23 ) vs @xmath105 and the average @xmath123 in eq.(18 ) vs @xmath0 are displayed for @xmath145 , and 2.4 . here ,
the orientation order develops continuously in a narrow temperature range , t_2 ( ) < t < t_1 ( ) , where @xmath146 and @xmath147 for @xmath78 . in our simulation , @xmath148 and @xmath149 increase with increasing @xmath36
. they are determined as crossover temperatures .
in this temperature window , a berezinskii - kosterlitz - thouless ( bkt ) phase @xcite is realized between the low - temperature ordered phase for @xmath150 and the high - temperature disordered phase for @xmath151 , where the orientation fluctuations are strongly enhanced at long wavelengths . in our model , each elliptic particle on a lattice point behaves as a rotator in the xy spin model under a symmetry - breaking free energy @xmath152 with @xmath153 , which arises from the underlying crystal structure @xcite . in accord with the theory @xcite , the structure factor @xmath144 in eq.(23 ) grows algebraically as s_q(k ) ~ k^-2 + ( k0.5 ) in the temperature range ( 24 ) with @xmath154 depending on @xmath0 ( @xmath155 at @xmath137 ) . as regards dynamics ,
the orientation fluctuations migrate in space on rather rapid time scales slightly below @xmath148 , but are frozen for @xmath156 ( see fig.13 below ) . considerably below @xmath149 ,
the three variants become distinct with sharp interfaces .
previously , for two - dimensional hard rods , bates and frenkel @xcite found a kosterlitz - thouless phase transition between the nematic phase and the isotropic phase for large aspect ratios and for low densities . as illustrated in the right panel of fig.2 , the orientation order induces lattice deformations .
in ordered states at low @xmath0 , each variant is composed of isosceles triangles elongated along its orientated direction parallel to one of the crystal axes . at low @xmath0 , their side lengths @xmath157 and @xmath1 are @xmath158 for @xmath78 and @xmath159 for @xmath79 under the periodic boundary condition , while we have @xmath160 at zero stress . thus this orientation transition is also a structural or martensitic one with spontaneous lattice deformations .
( top ) and strain @xmath161 ( bottom ) under fixed applied stress @xmath162 , 0.05 , 0.10 , and 0.15 in units of @xmath163 , where @xmath110 and @xmath78 .
the temperature @xmath0 was first decreased from 0.1 to 0.02 and it was then increased back to 0.1 , where @xmath164 .
hysteretic behavior appears between the cooling and heating paths . ] in eq.(28 ) divided by 4 ( left ) and the isobaric specific heat @xmath165 in eq.(32 ) ( right ) for @xmath110 on the cooling and heating paths in fig.4 in the nearly stress - free condition ( @xmath166 ) . here
@xmath167 is nearly equal to the effective shear modulus @xmath168 in eq.(29 ) .
softening against shear deformations and large energy fluctuations are conspicuous at the orientation transition . ] axis without impurities ( @xmath110 ) , where @xmath169 and @xmath78 .
left : strain @xmath170 vs applied stress @xmath171 in units of @xmath68 . for @xmath172 , there remains only the variant elongated along the @xmath173 axis .
after this cycle , the residual strain vanishes upon heating to @xmath174 .
right : fractions of the three variants during the cycle , which are stretched along the three crystal axes . ]
we have also performed simulation at a fixed stress @xcite , which allows an anisotropic shape change of the system at a structural phase transition . in figs.4 - 7
, we assumed a nos@xmath64-hoover thermostat @xcite and a parrinello - rahman barostat @xcite under the periodic boundary condition .
namely , we controlled the temperature @xmath0 and the stress along the @xmath173 axis written as _
a= , where @xmath175 represents the space average .
the @xmath173 axis is taken to be in the perpendicular direction in the figures .
hereafter , @xmath176 will be measured in units of @xmath68 .
when @xmath176 was held fixed at a positive value for a long time , a single - variant state elongated along the @xmath173 axis was eventually realized at low @xmath0 .
this was the case even for very small positive @xmath176 ( @xmath177 ) , since it serves as a symmetry - breaking field .
we also carried out many simulation runs exactly setting @xmath178 , where a few domains often remained in the final state depending on the initial conditions ( not shown in this paper ) .
when @xmath176 is controlled , the system length @xmath179 along the @xmath173 axis should be calculated . the strain @xmath161 is defined as = l_y
/ l_y0 - 1 , where @xmath180 is a reference system length to be specified below .
we may define young s modulus by e_e= 1/(/)_t . even in the nonlinear regime .
note that young s modulus is written as @xmath181 in the linear regime in terms of the bulk modulus @xmath53 and the shear modulus @xmath182 in two dimensions .
we may introduce the effective shear modulus @xmath168 replacing @xmath183 by @xmath184 as _ e = e_e/(4-e_e / k ) e_e/4 , where we have assumed @xmath185 .
substantial thermal hysteresis during cooling and heating has been observed in alloys around martensitic phase transitions @xcite .
et al_. also found thermal hysteresis numerically @xcite . in fig.4
, we show thermal hysteresis in our system for @xmath110 . that is , fixing @xmath176 , we decreased @xmath0 from 0.1 to 0.02 with a very slow cooling rate given by @xmath186 , where the variant elongated along the @xmath173 axis became dominant at low @xmath0 .
we then increased @xmath0 back to the initial high temperature with @xmath187 .
the curves of @xmath188 are those with a small symmetry breaking stress ( @xmath189 ) .
the reference length @xmath180 in eq.(27 ) is that at @xmath111 equal to @xmath190 , @xmath191 , @xmath192 , and @xmath193 for @xmath188 , 0.05 , @xmath194 , and 0.15 , respectively .
hysteretic behavior can be seen in the degree of orientation @xmath123 and the strain @xmath161 .
the width of the hysteresis loop is maximum for @xmath188 and shrinks to vanish for @xmath195 .
the transition at @xmath188 between the orientationally disordered and ordered states is shifted to lower temperatures by 0.01 than in the fixed - volume case in fig.1 .
see remark ( 4 ) in sec.vii for discussions on the stability of quasi - equilibrium states in fig.4 . when @xmath123 is small
, we may use the linear elasticity relations in two dimensions , & & k [ e_1-(t - t_0)]+ ( 2-e_1 ) = _ a , + & & k [ e_1-(t - t_0)]- ( 2-e_1 ) = 0 , where @xmath196 is the dilation strain , @xmath197 is the thermal expansion coefficient , and @xmath198 is a reference temperature . the small slope of the curves of @xmath161 at @xmath188 in the disordered regions in fig.4 arise from the thermal expansion . from these relations
we obtain e_1&= & _ a/2k + ( t - t_0 ) , + & = & ( k+)_a/4k+ ( t - t_0)/2 the data in fig.4 yield @xmath199 , @xmath200 , and @xmath201 with @xmath202 in the disordered phase . in the left panel of fig.5
, we show young s modulus @xmath184 in eq.(28 ) on the cooling and heating paths of @xmath188 in fig.4 . to calculate it
, we superimposed a small stress ( @xmath203 to the much smaller symmetry - breaking stress ( @xmath204 . remarkably , @xmath184 becomes very small around @xmath205 for @xmath110 @xcite .
similar minimum behavior of the shear modulus has been observed near the orientation and glass transitions @xcite , where the minimum depends on the mixture composition .
previously , using the correlation function expression , murat and kantor @xcite calculated the elastic constant to find its softening toward the orientation transition in two - dimensional ellipsoid systems .
nonlinear response behavior should appear even for very small applied strains near the transition .
additionally , in the right panel of fig.5 , we display the isobaric specific heat in the nearly stress - free condition ( @xmath166 ) along the cooling and heating paths expressed as c_p = /vt^2 , where @xmath206 is the total energy ( see eqs.(7 ) and ( 8) ) and @xmath207 is its deviation .
it is peaked at @xmath205 indicating enhancement of the energy fluctuations at the transition .
such specific heat anomaly has been measured near the orientation transition @xcite .
we also calculated the constant - volume specific heat @xmath208 using the data in fig.1 to find a similar peak around @xmath209 ( not shown in this paper ) .
next , we illustrate the shape memory effect taking place without dislocations . in fig.6 , we increased @xmath171 from 0 to 0.1 and then decreased @xmath171 back to 0 at @xmath169 , where @xmath210 with @xmath211 being on the stretching path and @xmath212 being on the return path . in this slow cycle ,
the system remained in quasi - static states . in the definition of @xmath161 in eq.(27 ) , @xmath180 is the initial system length @xmath213 . at @xmath214 , the fractions of the three variants
were nearly close to @xmath136 and one variant was elongated along the @xmath173 axis . in the very early stage @xmath215 ,
the system deformed elastically with @xmath216 .
however , in the next stage @xmath217 , the fraction of the favored variant increased up to unity with @xmath218 .
this inter - variant transformation occurred without dislocation formation . on the return path , the solid
was composed of the favored variant only with large @xmath219 .
as @xmath220 , there remained a remnant strain about 0.06 .
however , upon heating to @xmath174 above the transition , it disappeared and the solid again assumed a square shape .
we note that plastic deformations should occur at high strains . in the present simulation
, dislocations were indeed proliferated for @xmath221 ( or @xmath222 ) at @xmath169 .
in figs.7 - 11 , we treat mixtures of elliptic particles and large repulsive impurities . with increasing the impurity composition @xmath1 , the orientation disorder is enhanced and the long wavelength orientation fluctuations
are suppressed , resulting in `` orientation - strain glass '' . here , even for our anisotropic particle systems , we predict the nonlinear mechanical behavior studied for strain glass @xcite .
the bkt phase disappears with increasing @xmath1 . in the range @xmath107 ( left ) and order parameter amplitudes @xmath95 ( right ) with impurities ( black points ) at @xmath223 , @xmath194 and 0.2 , where @xmath224 and @xmath79 . ]
vs @xmath105 with @xmath79 and @xmath224 ( left ) and @xmath123 vs @xmath0 with @xmath79 ( right ) .
five curves correspond to @xmath225 , and 0.2 , which are gradually suppressed with increasing @xmath1 . ] around large impurities ( @xmath226 ) for @xmath227 and 0.2 , exhibiting planar anchoring of molecular orientations and clustering . ]
figures.7 - 9 are simulation results with a thermostat at fixed volume under the periodic boundary condition , where @xmath224 and @xmath79 .
the temperature was lowered from a high temperature as in the previous section .
the size ratio is fixed at @xmath228 . for @xmath229 ,
the system still forms a single hexagonal crystal with point defects at the impurity positions . in fig.7
, we present snapshots of @xmath16 and @xmath95 for three compositions as in fig.1 . in the top panel at @xmath223 ,
the impurities induce irregular orientation disorder , but not much affect the overall order such that large - scale domains are still distinct . in the lower panels with @xmath2300.1 and 0.2 ,
the orientation disorder increases and the domain sizes become finer . for @xmath231 ,
the system approaches orientation glass but with mesoscopically ordered regions still remaining . in fig.8
, increasing @xmath1 gives rise to suppression of @xmath144 at long wavelengths ( @xmath232 ) and @xmath123 in eq.(18 ) at low @xmath0 .
figure 9 displays expanded snapshots of the elliptic particles around the impurities .
we recognize that the alignments are mostly perpendicular to the surface normals , analogously to the parallel anchoring of liquid crystal molecules on the colloid surfaces @xcite .
moreover , we notice an apparent tendency of string - like clustering or aggregation of the impurities .
they tend to be localized along the interface regions between different variants , allowing formation of mesoscopically ordered regions of the elliptic particles even for @xmath233 . to examine the degree of clustering
, we may group the impurities into clusters .
let the two impurities @xmath43 and @xmath44 belong to the same cluster if their distance is shorter than @xmath234 .
then we obtain the numbers @xmath235 of the @xmath236 clusters ( those consisting of @xmath236 impurities ) , where @xmath237 and @xmath238 .
the probability that one impurity belongs to one of the @xmath236 clusters is @xmath239 .
the average cluster size is defined as _ cl= _
p_cl()= _ ^2 n_cl()/n_2 in fig.7 , we have @xmath240 , 2.04 , and 4.79 for @xmath223 , 0.1 , and @xmath241 , respectively .
axis with impurities , where @xmath224 , @xmath233 , and @xmath79 .
top : @xmath170 vs @xmath171 ( left ) and @xmath123 vs @xmath171 ( right ) . for @xmath242 , @xmath161 and @xmath123
increase steeply . for @xmath243 , there remains only the variant elongated along the @xmath173 axis .
after this cycle , the residual strain is 0.04 , which vanishes upon heating to @xmath174 .
bottom : snapshots of @xmath16 for @xmath244 , and 0.15 in the transition region , where large - scale orientation fluctuations can be seen but there is no dislocation . ] we also observed a shape - memory effect even in orientation glass , where small disfavored domains were gradually replaced by favored ones upon stretching . in this effect , no dislocation was formed at low @xmath0 . in fig.10 , at @xmath224 , we increased @xmath176 slowly at @xmath245 from 0 up to 0.2 , where the variant elongated along the @xmath173 axis becomes increasingly dominant .
we then decreased @xmath176 back to 0 at @xmath246 . between these two paths ,
significant differences can be seen in the degree of orientation @xmath123 and the strain @xmath161 .
the @xmath161 is given by eq.(27 ) with @xmath180 being the initial system length at @xmath224 and @xmath188 .
on the stretching path , there appear four stress ranges : @xmath247 for @xmath248 , @xmath249 for @xmath250 , @xmath251 for @xmath252 .
remarkably , the response is elastic in the first range and is very large with @xmath161 increasing steeply from @xmath253 to @xmath254 in the third range . for @xmath195 and on the return path
, @xmath168 is of order unity and we can see considerable variations in @xmath161 and @xmath123 , where the fractions of the disfavored variants significantly change around the impurities .
in contrast , in the one - component case in fig.6 , we have found no such changes once a single - variant state is realized . in the bottom panels of fig.10 , we display snapshots of @xmath16 at four points a , b , c , and d where @xmath255 , 1.5 , and 0 , respectively .
see the bottom left panel of fig.7 for the snapshot at the initial time in the same run . between a and b , the orientation and the strain increase abruptly .
in this transition region , we notice emergence of large - scale orientation fluctuations taking stripe shapes and making angles of @xmath256 with respect to the @xmath257 axis . in stress and thermal cycles in glass ,
the impurities pin the orientation fluctuations in quasi - stationary states under very slow time variations of @xmath176 and @xmath0 , yielding the history - dependence of the physical quantities .
so far , the crystal structure has been little affected by the orientation fluctuations at @xmath228 for @xmath227 and 0.2 . however , if we adopt a larger size ratio and/or a larger composition , the positional ( structural ) disorder is increasingly enhanced , resulting in usual positional polycrystal or glass . in our case ,
the orientation disorder is more enhanced than the positional disorder .
this is in sharp contrast to liquid crystal systems where the nematic order precedes the crystallization . in fig.11
, we set @xmath258 and @xmath79 to obtain polycrystal for @xmath227 and 0.2 . in the left , the orientation angles @xmath259 are displayed , where there still remains noticeable mesoscopic orientation order . in the left , sixfold bond - orientation ( crystal ) angles
@xmath260 are displayed , where we introduce @xmath260 for each elliptic particle @xmath44 in the range @xmath261 by @xcite _
k = z_j , here , @xmath262 is the angle of the relative position vector @xmath263 with respect to the @xmath257 axis , the particle @xmath105 is within the range @xmath264 , and @xmath265 and @xmath266 are the absolute value and the phase angle of the left hand side , respectively . for @xmath227 ,
one large grain is embedded in a crystal containing many point defects , where angle differences are of order @xmath267 degrees .
on the other hand , for @xmath233 , many grains appear with much larger angle differences . in the range @xmath107 ( left ) and sixfold bond orientation angle @xmath268 in eq.(34 ) in the range @xmath269 ( right ) in polycrystal states for @xmath227 ( top ) and 0.2 ( bottom ) , where @xmath79 and @xmath224 .
the size ratio is increased to @xmath258 . ]
we now discuss the rotation dynamics of the elliptic particles@xcite . in two dimensions ,
we consider the time - dependent angle - distribution function defined by g(t , ) = _ 1jn_1 , where the average @xmath175 is taken over the initial time @xmath270 and over several runs . here , @xmath271 tends to @xmath272 as @xmath273 and is broadened for @xmath274 . in particular , we treat the first two moments @xmath275 and @xmath276 written as g_1(t)&= & _ 0 ^ 2d g(t , ) ( ) , + g_2(t)&= & _ 0 ^ 2 d g(t , ) ( 2 ) . since these two functions are unity as @xmath277 , we introduce two relaxation times , @xmath278 and @xmath279 , by g_1(_1 ) = e^-1 , g_2(_2 ) = e^-1 .
these times grow as @xmath0 is lowered .
we plot @xmath275 and @xmath276 vs @xmath280 in fig.12 and @xmath278 vs @xmath0 in fig.13 .
( left ) and @xmath276 ( right ) in eqs.(36 ) and ( 37 ) for @xmath110 and @xmath78 ( top ) and for @xmath233 , @xmath228 , and @xmath79 ( bottom ) .
relaxations are slowed down as @xmath0 is lowered .
the @xmath275 decays due to turnover motions , while @xmath276 due to configuration changes . for @xmath281
, @xmath276 tends to a finite constant as @xmath282 . ] from @xmath275 for ( a ) @xmath110 and @xmath78 , ( b ) @xmath227 and @xmath78 , ( c ) @xmath110 and @xmath79 , and ( d ) @xmath227 and @xmath79 .
it is the time scale of successive turnover motions . for ( a ) and ( c )
, @xmath278 grows steeply in the bkt phase ( @xmath283 ) . for strain glass ( b ) and
( d ) , the bkt phase is nonexistent and @xmath278 grows as @xmath0 is lowered . ] as a marked feature , the elliptic particles sometimes undergo the turnover motion @xmath284 or @xmath285 taking place in a microscopic time @xmath286)@xcite . in terms of the orientation vector @xmath35 , we also have g_1(t ) = _
1j n_1 /n_1 , so that the times between successive turnovers of an elliptic particle are of order @xmath278 in eq.(38 ) . on the average over all the elliptic particles ,
the turnover motions give rise to a peak in @xmath287 at @xmath288 . in our simulation , it is nearly of the gaussian form for @xmath289 in the range @xmath290 as g(t , ) , where the variance @xmath291 is a constant about @xmath292 in the present case .
the integral of this gaussian peak is equal to the coefficient @xmath293 , so @xmath293 has the meaning of the turnover probability per elliptic particle in the time interval @xmath294 $ ] . in terms of @xmath278 , we find the linear growth , a(t ) c_1 t/_1 , in the early time range @xmath295 . in our system @xmath296 .
on the other hand , @xmath276 is unchanged by the instantaneous turnover motions , so it relaxes due to the orientational configuration changes involving the surrounding particles .
we found the inequality @xmath297 at any @xmath0 and @xmath1 in our simulation . in fig.12 ,
both @xmath275 and @xmath276 relax considerably in the early time region @xmath298 due to the thermal rapid motions of the orientations without configuration changes .
for @xmath299 the fitting @xmath300 $ ] fairly holds , where @xmath301 decreases from unity to about 0.5 as @xmath0 is lowered .
furthermore , for @xmath281 , @xmath276 tends to a nonvanishing positive constant @xmath302 at large @xmath280 @xcite . in our case
, this plateau appears because the anchoring of the elliptic particles around the impurities becomes nearly permanent at low @xmath0 .
thus we found that the plateau height @xmath302 increases with lowering @xmath0 and with increasing @xmath1 . in fig.13 ,
the two curves for @xmath110 indicate that @xmath278 is short ( @xmath303 for @xmath304 , increases steeply in the bkt region @xmath305 , and grows further in the ordered region @xmath306 in the thermal activation form , _ 1 ~(t_0/t ) ( t t_2 ) .
we have @xmath307 at @xmath78 on curve ( a ) and @xmath308 at @xmath309 on curve ( c ) .
in addition , @xmath310 for @xmath311 but @xmath312 for @xmath156 .
in fact , for @xmath78 , @xmath313 is about @xmath314 at @xmath315 and is about @xmath316 at @xmath317 . for @xmath281
, the turnover motions still occur with @xmath318 .
however , in fig.13 , the relaxation behavior for @xmath281 is very different from that for @xmath110 . in the disordered phase with @xmath304 , @xmath278 for @xmath281
is longer than @xmath278 for @xmath110 due to the impurity pinning . for @xmath156 , on the contrary
, @xmath278 for @xmath281 is shorter than @xmath278 for @xmath110 .
that is , the turnover motions are more frequent in orientation glass with @xmath281 than in the orientationally ordered phase with @xmath110 , as ought to be the case . above @xmath149 , the impurity anchoring gradually becomes transient .
it is worth noting that chong _
et al _ @xcite studied the orientation dynamics of a glass - forming binary mixture of dumbbells using the angle relaxation functions @xmath319 in three dimensional molecular dynamics simulation , where @xmath320 is the legendre polynomial of order @xmath236 and @xmath35 is the orientation vector of particle @xmath44 .
the relaxations of @xmath321 and @xmath322 for small dumbbell anisotropy in their paper closely resemble those of @xmath275 and @xmath276 for @xmath233 in fig.12 .
( left ) and sixfold crystal angles @xmath268 with addition of small attractive impurities with @xmath227 ( top ) , 0.2 ( middle ) , and 0.3 ( bottom ) , where @xmath174 , @xmath323 @xmath324 , and @xmath325 .
the orientation disorder is stronger than the positional disorder .
the cooling rate from @xmath326 to 0.1 is @xmath327 . ]
in this section , we further treat another intriguing case of small attractive impurities with @xmath324 in eq.(3 ) and with @xmath325 in eq.(6 ) .
such small impurities tend to be expelled from the ordered domains of the host particles .
we shall see that they form clusters . though not shown in this paper
, we performed simulation runs for small repulsive impurities with @xmath324 and @xmath328 , where most of the impurity aggregates are stringlike and the anchoring of the elliptic particles is planar .
however , if the anisotropy strength of attraction @xmath49 is increased at fixed @xmath36 , the aggregates becomes increasingly compact . for @xmath329
, the aggregates can `` solvate '' several elliptic particles in the homeotropic alignment@xcite . with further increasing @xmath49
, even a single impurity creates a solvation shell composed of several elliptic particles like a small metallic ion in water .
( left ) and the sixfold crystal angles @xmath260 ( right ) from the same data .
the domains ( left ) are finer than the grains ( right ) .
the cooling rate from @xmath326 to 0.1 is @xmath327 . ] in fig.14 , we show snapshots of @xmath259 and @xmath260 of all the particles , where @xmath174 , @xmath323 @xmath324 , and @xmath325 . here , we set @xmath327 . for @xmath227 ,
the system is still in a single crystal state , but the orientational domain structure induces large - scale elastic deformations , leading to close resemblance of the patterns of @xmath259 and @xmath260 . for @xmath233 ,
the orientational domains are much finer and a polycrystal state is realized with larger grains ( @xmath330 ) . for @xmath331 ,
the orientation order is much more suppressed and a positional glass state is realized with mesoscopic heterogeneities still remaining . in fig.15
, we display expanded snapshots of @xmath259 ( left ) and @xmath260 in eq.(34 ) ( right ) in the box regions in fig.14 .
the alignments of the elliptic particles around the impurities are mostly parallel to the surface normals .
this is analogous to the homeotropic anchoring of liquid crystal molecules on the colloid surfaces @xcite .
we notice a tendency of clustering or aggregation of the impurities . comparing the left and right panels , we recognize that the interfaces are finer than the grain boundaries .
that is , the interfaces can be seen both on the grain boundaries and within the grains .
the impurities tend to be localized on the interface regions between different variants .
( left ) and sixfold crystal angles @xmath268 ( right ) with addition of small attractive impurities with @xmath227 .
the parameter values are common to those in the top panel of fig.14 , but the cooling rate from @xmath326 to 0.1 is @xmath332 . here , the degree of clustering is weaker , resulting in a polycrystal state . ] of an impurity belonging to @xmath236 clusters , where @xmath333 is the cluster number composed of @xmath236 impurities . here
small attractive impurities are considered for slow cooling in fig.14 and fast cooling in fig.16 .
distribution is broader for slower cooling . ]
the degree of impurity clustering should be decreased with increasing the cooling rate @xmath334 for long diffusion times of impurities . in fig.16 , @xmath334 is @xmath335 and is 500 times faster than in fig.14 , where the other parameters are common .
we give snapshots of @xmath259 and @xmath260 at @xmath227 , where the clustering can be more evidently seen than for @xmath233 and 0.3 .
while a single crystal has been realized in the top panel of fig.14 , a polycrystal state is realized with large angle differences in fig.16 .
let the two small impurities @xmath43 and @xmath44 belong to the same cluster if their distance is shorter than @xmath336 . then we obtain the number @xmath337 of clusters composed of @xmath236 impurities .
in fig.17 , we show the cluster size distribution @xmath338 ( @xmath339 ) for the examples in figs.14 and 16 .
the average cluster size @xmath340 in eq.(33 ) increases with @xmath1 as @xmath341 for @xmath227 , 3.43 for @xmath233 , and 4.61 for @xmath342 under the slow cooling in fig.14 , while @xmath343 for @xmath227 under the fast cooling in fig.16 .
it is known @xcite that water becomes glass at low @xmath0 with addition of a considerable amount of licl , where small hydrophilic li@xmath344 and cl@xmath345 ions solvate several water molecules via the strong ion - dipole interaction .
the resultant orientation anchoring of water molecules should even prevent formation of the crystal order at high salt concentrations , resulting in the observed positional glass .
it is natural that the cooling rate influences the degrees of ion clustering and vitrification .
we have presented an angle - dependent lennard - jones potential for elliptic particles and impurities , which depends on the orientation angles of the interacting particles . using this potential ,
we have performed simulation of 4096 particles on very long time scales ( @xmath346 ) in two dimensions .
our main results are as follows .
+ ( i ) in sec.ii , we have presented our model potential , where the anisotropy strengths are characterized by @xmath36 for the repulsive part in eq.(5 ) and @xmath49 for the attractive part in eq.(6 ) .
the aspect ratio of the elliptic particles is given by @xmath347 .
in this paper , @xmath36 is of order unity , so we have assumed weak particle anisotropy to find crystallization at a high temperature above the orientation transition .
+ ( ii)in sec.iii , we have presented simulation results for one - component systems of elliptic particles by changing the temperature @xmath0 to produce figs.1 - 6 .
the domain patterns in fig.1 at low @xmath0 are those observed on hexagonal planes . in our case , the berezinskii - kosterlitz - thouless phase@xcite is realized in a temperature window , where the orientation fluctuations are much enhanced at long wavelengths as indicated by the structure factor @xmath144 in fig.3 .
we have shown thermal hysteresis in fig.4 , singular behaviors of the shear modulus and the specific heat in fig.5 , and a shape - memory effect in fig.6 . + ( iii ) in sec.iv , we have examined orientation - strain glass of elliptic particles and large repulsive impurities with the size ratio @xmath228 in figs.7 - 10 .
the orientations of the elliptic particles are pinned at the impurity surfaces in the planar alignment in fig.9 . the shape - memory effect in strain glass
is marked in fig.10 .
positional disorder also emerges for @xmath258 in fig.11 .
+ ( iv ) in sec.v , we have studied the rotational dynamics of the elliptic particles . in fig.12
, @xmath275 decays due to the turnover motions of the elliptic particles , while @xmath276 decays due to the configuration changes . in fig.13 ,
the turnover relaxation time @xmath278 grows at low @xmath0 and behaves differently with and without impurities .
+ ( v ) in sec.vi , we have examined the effect of small attractive impurities on the orientation disorder and the positional disorder in fig.14 .
the impurity effect is stronger on the former than on the the latter .
the elliptic particles are homeotropically anchored at the impurity surfaces in fig.15 .
the clustering of impurities is suppressed for rapid cooling as in figs.16 and 17 .
+ we further make critical remarks as follows : + ( 1 ) in our simulation , we used a nos@xmath64-hoover thermostat ( nht ) @xcite in all the figures and a nht and a parrinello - rahman barostat @xcite in figs.4 - 6 , and 10 . in future work , we should examine the coupled dynamics of the translational and orientational degrees of freedom @xcite without thermostats and barostats in the system interior . + ( 2 )
there has been no systematic measurement of the mechanical properties of orientationally ordered , multi - variant crystal and orientation glass .
such experimental results could be compared with those from shape - memory alloys @xcite .
weak elasticity was observed in orientationally disordered solids above the transition ( called `` plastic solids '' ) in creep experiments @xcite . also , as far as the authors are aware , there has been no experimental information of the impurity clustering in any physical systems exhibiting mesoscopic heterogeneities .
+ ( 3 ) in this paper , the particle anisotropy is not large , which favors formation of crystal order . for large anisotropy
, liquid crystal phases should appear @xcite , where the impurity effect is of great interest .
as suggested by the experiment @xcite , addition of a considerable amount of impurities leads to the orientation order only on mesoscopic scales in liquid crystal phases . in such states
, we expect large response to applied electric field .
+ ( 4 ) in figs.1 and 4 , our system undergoes a structural phase transition gradually in a narrow temperature window even for the one - component case . in our model
, a gradual phase transition still occurs in the stress - free condition without impurities .
however , we also stopped the cooling and waited for a long time ( @xmath348 ) at @xmath205 on the stress - free cooling path in fig.4 ; then , we observed a transition to the ordered single - variant phase ( not shown in this paper ) .
thus , in future work , we need to calculate the gibbs or helmholtz free energy to decide whether the system is in equilibrium or in a metastable state .
+ ( 5 ) as well as the orientation fluctuations , the displacement fluctuations are also enhanced around the orientation transition , as indicated by fig.5 and by the previous experiments @xcite .
in addition , according to cowley s classification of elastic instabilities @xcite , our phase transitions belong to type - i instabilities where acoustic modes become soft in particular wave vector directions
. + ( 6 ) the disordering effect induced by impurities prevents a sharp transition @xcite .
thus there is no sharp phase boundary between the high - temperature orientationally disordered phase and the low - temperature orientation - strain glass phase .
these two phases change over gradually with varying @xmath0 as in the cases of positional glass transitions .
+ ( 7 ) ding _ et al . _
@xcite numerically studied the superelasticity , which arises from a stress - induced martensitic phase transition @xcite .
we also realized this phenomenon for anisotropic particles , which will be reported elsewhere .
+ ( 8) we will also report three - dimensional simulation on mixtures of spheroidal particles and spherical ones without and with the dipolar interaction .
we shall see finely divided domains produced by impurities and large responses to applied strain and electric field .
this work was supported by grant - in - aid for scientific research from the ministry of education , culture , sports , science and technology of japan .
the authors would like to thank takeshi kawasaki , osamu yamamuro , hajime tanaka , and hartmut l@xmath349wen for informative discussions .
the numerical calculations were carried out on sr16000 at yitp in kyoto university .
e. s. matsuo , m. orkisz , s .-
sun , y. li and t. tanaka , macromolecules , * 27 * , 6791 ( 1994 ) ; f. ikkai and m. shibayama , phys .
* 82 * , 4946 ( 1999 ) .
e. mendes , r. oeser , c. hayes , f. bou and j. bastide , macromolecules * 29 * , 5574 ( 1996 ) .
j. k@xmath351pfer and h. finkelmann , macromol .
* 195 * , 1353 ( 1994 ) ; y. yusuf , j .- h . huh , p. e. cladis , h. r. brand , h. finkelmann , and s. kai , phys . rev .
e * 71 * , 061702 ( 2005 ) ; k. urayama , e. kohmon , m. kojima , and t. takigawa , macromolecules * 42 * , 4084 ( 2009 ) .
s. sacanna and d. j. pine , current opinion in colloid @xmath350 interface science * 16 * , 96 ( 2011 ) ; a. f. demirrs , p. m. johnson , c. m. van kats , a. van blaaderen , and a. imhof , langmuir * 26 * , 14466 ( 2010 ) .
z. zheng , f. wang , and y. han , phys .
* 107 * , 065702 ( 2011 ) .
k. h. michel , phys .
lett . * 57 * , 2188 ( 1986 ) . v. westphal , w. kleemann , and m. d. glinchuk , phys .
. lett . * 68 * , 847 ( 1992 ) .
r. vasseur and t. lookman , phys .
b * 81 * , 094107 ( 2010 ) ; n. shankaraiah , k. p. n. murthy , t. lookman , and s. r. shenoy , phys . rev .
b * 84 * , 064119 ( 2011 ) . c. a. angell , chem . rev . * 102 * , 2627 ( 2002 ) ; b. prvel , j. f. jal , j. dupuy - philon , and a. k. soper , j. chem .
phys . * 103 * , 1886 ( 1995 ) ; m. kobayashi and h. tanaka , phys .
rev . lett . * 106 * , 125703 ( 2011 ) .
g. de gennes and j. prost , _ the physics of liquid crystals _ ( clarendon , oxford , 1993 ) . j. g. gay and b. j. berne , j. chem .
phys . * 74 * , 3316 ( 1981 ) ; j. t. brown , m. p. allen , e. m. del rio , and e. miguel , phys .
e * 57 * , 6685 ( 1998 ) .
h. shintani and h. tanaka , nat .
* 2 * , 200 ( 2006 ) . c. vega and p. a. monson , j. chem .
* 107 * , 2696 ( 1997 ) ; c. de michele , r. schilling , and f. sciortino , phys .
* 98 * , 265702 ( 2007 ) ; m. marechal and m. dijkstra , phys .
e * 77 * , 061405 ( 2008 ) ; m. radu , p. pfleiderer , and t. schilling , j. chem . phys . * 131 * , 164513 ( 2009 ) .
r. sinclair and j. dutkiewicz , acta metell .
* 25 * , 235 ( 1977 ) ; y. kitano , k. kifune , and y. komura , j. phys .
( paris ) * 49 * , c5 - 201 ( 1988 ) ; c. manolikas and s. amelinckx , phys .
( a ) * 60 * , 607 ( 1980 ) ; _ ibid_. * 61 * , 179 ( 1980 ) .
|
using a modified lennard - jones model for elliptic particles and spherical impurities , we present results of molecular dynamics simulation in two dimensions . in one - component systems of elliptic particles , we find an orientation phase transition on a hexagonal lattice as the temperature @xmath0 is lowered .
it is also a structural one because of spontaneous strain . at low @xmath0 ,
there arise three martensitic variants due to the underlying lattice , leading to a shape memory effect without dislocation formation .
thermal hysteresis , a minimum of the shear modulus , and a maximum of the specific heat are also found with varying @xmath0 . with increasing the composition @xmath1 of impurities ,
the three kinds of orientation domains are finely divided , yielding orientation - strain glass with mesoscopically ordered regions still surviving .
if the impurities are large and repulsive , planar anchoring of the elliptic particles occurs around the impurity surfaces . if they are small and attractive , homeotropic anchoring occurs .
clustering of impurities is conspicuous . with increasing the anchoring power and/or the composition of the impurities
, positional disorder can also be enhanced .
we also investigate the rotational dynamics of the molecular orientations .
|
supernovae ( sne ) and supernova remnants ( snr s ) are thought to be the main `` drivers '' of the starburst phenomenon ( @xcite ) .
detailed radio observations for supernovae and supernova remnants are few , due to the fact that only a small subset of the extragalactic sne discovered each year produce detectable radio emission .
those events for which monitoring data is available ( e.g. @xcite ) have revealed considerable information regarding the evolution of the properties of the expanding shock wave and its emission processes . in intense starbursts like m82 , where a significant population of massive stars has had sufficient time to evolve to the sn stage , we have the unique opportunity to study a collection of snr s of similar age and origin , and their interaction with the surrounding environment .
the observations are briefly outlined in section 2 .
we discuss the technique used to fit spectral models in section 3 .
then in section 4 we discuss the physical implications of our modelling . throughout the paper ,
we assume that the distance to m82 is 3.63 mpc ( @xcite , @xcite ) , so that 1@xmath30 corresponds to 17.6 pc .
we have used observations of m82 at @xmath0 20 , 6 , 3.6 , 2 , and 1.3 cm . to supplement these observations ,
we have added published flux densities at @xmath0 74 and 18 cm ( 408 and 1666 mhz ) for individual radio point sources as measured with the merlin interferometer by @xcite and @xcite .
the @xmath0 2 and 1.3 cm data were first presented in @xcite and are composed of vla data collected at all four configurations , including data from the flux monitoring program of kronberg & sramek ( see e.g. kronberg & sramek 1992 ) between 1983 and 1992 .
the @xmath4 20 cm data is from two a - array observations ( @xcite ) .
the @xmath4 6 cm data is a combination of vla data from its a- , c- and d - configurations .
the @xmath4 3.6 cm data were collected in 1994 - 5 in the vla s a- , b- and c - configurations , and were combined with a previous observation in the d - configuration , first presented in @xcite .
the combined datasets at @xmath020 , 6 , 3.6 , 2 and 1.3 cm were mapped and clean - deconvolved using the nrao s aips imaging software . to eliminate the bright background emission and
concentrate on the point sources , we restricted the lower @xmath5-baseline range so that no structures spatially larger than 4@xmath6 were present .
this range proved most effective - restricting the data to smaller structures led to point spread functions that had large negative sidelobes , that in turn led to deconvolution problems ; allowing larger structures led to a non - zero background that tended to cause problems with the determination of the integrated fluxes of the point sources .
the complete list of images used is given in table 1 .
a full description of the data reduction of the combined vla datasets will be given in a forthcoming paper , where we will also present a study of the diffuse radio emission defining the nuclear region of m82 .
the datasets were mapped at 0@xmath330 and 0@xmath320 resolution .
these beamsizes were chosen for proper flux density scaling with the lower frequency data of unger _
_ , which has a fwhm of 0@xmath325 at 18 cm .
at the same time we wanted to guard against missing flux from possible partial resolution of the radio sources . to each source
we fitted 2-d gaussian profiles using the aips task imfit . for a given wavelength
, we then compared the results of the fitting at each resolution against each other , and against the result of adding up the flux in each pixel of the radio source ( aips task tvstat ) . for this latter pixel - adding method
, we only included those pixels of the radio source that contained emission above the background local to the source .
this background was determined by eye , by examining profiles of slices through the sources .
fluxes determined in this way served as a quick check on the gaussian fits , and were not used as data points in any of our spectral fits . in general , the values for the integrated fluxes determined using each method agreed well with each other , although the pixel - added tvstat fluxes were generally smaller .
the flux values quoted in table 2 are the weighted average of the gaussian - fitted fluxes obtained at both resolutions , except for sources that were partially resolved at 0@xmath320 , for which we used only the 0@xmath330 results .
these latter were unresolved in all cases .
the errors given for each fit are the larger of the errors calculated by imfit , and the 3-@xmath7 level on each map . for the sources in m82 , these resolutions are ideal : they are sufficiently high to isolate the sources from the background emission , and yet not so high as to resolve the sources out .
the best resolution obtainable for the @xmath4 20 cm data was a 1@xmath315@xmath80@xmath392 beam in position angle 38@xmath96 ( ccw from north ) , which matches the resolution of the @xmath4 74 cm data .
we chose only those point sources that were clearly separated from any neighbours .
we compared our @xmath4 3.6 cm results against those of @xcite , who had a resolution of 0@xmath3182 .
our fluxes are systematically greater than theirs . on average , the values of @xcite
are only 66% of our values .
we speculate that the difference is due to a combination of ( 1 ) missing flux on their part due to partial resolution of the sources , and ( 2 ) possible inclusion on our part of contaminating larger - scale flux along the same line of sight to the source .
since we intentionally excluded structures larger than 4@xmath6 , this presumed large - scale flux would be of very low surface - brightness and would be physically unrelated to the compact radio sources .
the values listed in this publication are consistent with a power - law interpretation of the high - frequency shape of the spectra of the point sources .
the complete results for the fluxes of 25 radio point sources are given in table 2 ( we treat 41.9 + 58 separately , in table 3 ) .
of the 19 sources listed by @xcite , we found that one of their sources ( 47.7 + 62 ) had no corresponding feature in our maps , and two of their sources could not be identified uniquely with a particular source at our wavebands . on the other hand
, we were able to identify every source listed by @xcite .
[ [ the - source-41.958 ] ] the source 41.9 + 58 + + + + + + + + + + + + + + + + + + to correct for the decaying flux of 41.9 + 58 , we used the variability results from @xcite to determine that the approximate decay rate is 8.8%/year .
we applied this rate to the fluxes measured by wills _
et al . _ . our chosen epoch for the maps
is may 1993 ( `` 1993.4 '' ) .
the flux values used for the spectral fit are listed in table 3 .
we note that the corrected epoch-1984 408 mhz flux is 72@xmath108 mjy , whereas the corrected epoch-1997 flux is 117@xmath108 mjy .
this discrepancy can be understood by realizing that for this source , @xcite discovered that the turnover frequency is evolving over time , as is the spectral index at all frequencies .
in particular , the source is becoming optically thin at progressively lower frequencies .
this effect has also been seen in other young radio supernovae ( @xcite , @xcite ) .
the measured decay rate of 8.8%/year is derived from @xmath4 6 cm ( 4.8 ghz ) data .
therefore our flux corrections are most accurate within the powerlaw regime of the spectrum , i.e. the higher frequencies , where the spectral index has been evolving slowly .
our decay estimate is least certain at the lower frequencies where the emission is optically thick .
overall we are probably underestimating the errors associated with the flux of 41.9 + 58 .
to each point source , we fitted two composite models to the emission spectrum : ( 1 ) a model of synchrotron and thermal ( ) emission , with a thermally absorbing screen located between the source and the observer , and ( 2 ) a co - spatial mix of synchrotron and thermally - absorbing gas .
these two models are represented by the equations ( cf .
@xcite ) : @xmath11 \;\;\;\;(screen)\\
i(\nu ) & = & \left [ i_{sy } \nu^{2.1+\alpha } + i_{th } \left ( \frac{\nu}{\nu_{\tau=1 } } \right)^2 \right ] \left [ 1 - e^{-(\nu/\nu_{\tau=1})^{-2.1 } } \right ] \;\;\;\;(mixed)\end{aligned}\ ] ] in the above expressions , @xmath12 and @xmath13 are related in a simple way to the optically thin synchrotron and thermal fluxes respectively , @xmath14 is the ( optically thin ) non - thermal spectral index ( defined so that @xmath15 ) , and @xmath16 is the frequency at which the optical depth to thermal absorption becomes unity . since the number of data points and free parameters are comparable , standard least - squares fitting procedures were inadequate for finding the absolute minimum of the chi squared hypersurface .
therefore , we wrote out the @xmath17 equations and their derivatives in full .
it was discovered that two of the parameters , ( @xmath12 and @xmath13 ) could be expressed completely in terms of the other two ( @xmath14 and @xmath16 ) .
a grid over reasonable values of these two latter parameters was constructed , and a value for @xmath17 at each point was generated .
we chose the set of ( @xmath14 , @xmath16 ) that yielded the lowest value for @xmath17 .
the spectral `` best '' fits are included in tables 4 , 5 and 6 , and are plotted in figure 1 . in table 4 , we list the best fit parameters for those sources that have no firm evidence for absorption .
tables 5 and 6 list the best fit parameters for absorbed sources - for table 5 we have used a screen model fit ( equation 1 ) , and for table 6 we have used a mixed model fit ( equation 2 ) .
only one source had a very uncertain spectral shape , 47.37 + 680 .
for this source we plotted both the power - law fit and a fit for the screen model ; for the mixed model fit no sensible solution was reached . to evaluate the reliability of our fits ,
we generated test data , and examined slices through the @xmath17 surface . the surfaces
were irregularly shaped , with a pronounced absolute minimum at the location expected for the test data . by varying the spectral index @xmath14 between -0.11 and -2.20 in increments of 0.01 , and the optical thickness parameter @xmath16 from 0.10 to 4.00 ghz in increments of 0.01 ghz
, we were able to unambiguously identify the best fit solution within an error equal to the size of the grid increments .
a separate test on each source was applied to see if the spectrum could be purely thermal ( @xmath18 ) .
[ [ magnetic - field - associated - pressures ] ] magnetic field , associated pressures + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + fully 19 of the 26 sources show optically thin , non - thermal spectral indices of @xmath14 @xmath19 -0.49 . for those sources we calculated the equipartition magnetic field strength ( cf .
this value is derived from the calculated minimum energy in particles and fields required to produce the observed radio signature , within an estimated radiating volume .
the values scale with the undetermined parameters @xmath20 , which is the ratio of the relativistic proton to electron energy densities , and @xmath21 , which is the fraction of the source s volume occupied by relativistic particles and fields ( refer to the notes in tables 4 , 5 & 6 ) . from direct cosmic ray observations near the solar system
, it is thought that @xmath20 is between 40 - 100 in the milky way ( see e.g. @xcite ) .
we assumed that the intrinsic spectral index for each source was constant from 0.01 to 100 ghz . in most of the sources ,
the equipartition magnetic field is about 1 - 2 mgauss .
bright , compact radio sources in other galaxies have similar inferred field strengths ( @xcite ) .
these values are higher than but comparable to the inferred ism magnetic field strength for m82 of @xmath22 50 - 100 @xmath23 g ( @xcite , @xcite , @xcite ) , and are much higher than the measured value for the magnetic field in our galaxy , @xmath22 3 @xmath23gauss ( @xcite ) .
the corresponding minimum pressures of the synchrotron plasma within the point sources have been calculated using equation ( 7.15 ) of @xcite .
they are of order @xmath22 10@xmath24 dynes @xmath25 .
these values are derived assuming that the sources have particles and fields distributed homogeneously throughout a spherical volume , and hence represent lower limits .
the total pressure of the ism in the nucleus of m82 has been estimated by @xcite .
they derive values @xmath22 10@xmath24 @xmath26 10@xmath27 dynes @xmath28 .
these values would suggest that the brighter point sources ( at least ) in this survey are expanding into a lower - pressure medium .
all of these values are far higher than the pressure of the ism of the galaxy , which is @xmath22 10@xmath29 dynes @xmath25 ( @xcite ) .
this evidence supports the theory that supernovae add mechanical and magnetic energy into the ism of starburst galaxies .
the spectral fits for the steep - spectrum sources ( @xmath14 @xmath19 -0.49 ) show very low levels of free - free emission .
these values are generally so low as to have a negligible effect on the actual shape of the spectra .
the exceptions are sources 43.31 + 591 , 46.52 + 638 , 46.56 + 738 , and 46.70 + 670 .
[ [ low - frequency - turnovers ] ] low frequency turnovers + + + + + + + + + + + + + + + + + + + + + + + the most striking feature of the spectra is that 17 of the 20 steep - spectrum sources ( and one flat - spectrum source , see below ) show a low - frequency turnover .
we will consider the nature of this phenomenon in this section .
the discussion applies to homogeneous sources .
a turnover at the lower end of a synchrotron spectrum can result from many situations : the synchrotron source can become optically thick due to self - absorption , the injected relativistic electrons can have a low - energy cutoff , plasma effects , or absorption by intervening .
we will show that intervening is the most likely cause .
optically thick synchrotron emission has a spectral index @xmath30 , where @xmath31 .
synchrotron self - absorption becomes important for sources with a brightness temperature @xmath32 k. the brightest source in our sample ( 41.9 + 58 ) has @xmath33 k at 1.4 ghz , well below the synchrotron self - absorption threshold , so we can discount self - absorption because the surface brightnesses are too low .
if the synchrotron radiation is emitted in a plasma , then the presence of thermal electrons will decrease the emission efficiency of the relativistic electrons , called the tsytovich - razin effect ( @xcite , @xcite ) .
this effect becomes important at frequencies @xmath34 ( cgs ) . for our derived @xmath35 mg ,
this effect becomes noticeable at 1 ghz for an electron number density of @xmath36 @xmath37 .
at such high densities , however , free - free absorption is a more important effect .
the free - free optical depth is @xmath38 at n@xmath39 = 5 @xmath8 10@xmath40 @xmath37 , t@xmath39 = 10@xmath40 k , @xmath41 = 1 ghz , and for a typical m82 snr of radius s = 0.8 pc ( 0.05 ) , the free - free optical depth is 660 !
we can thus discount a tsytovich - razin - like cutoff to explain the low - frequency shape of the spectra .
if the source is optically thin at all frequencies , then the spectrum will have a spectral index @xmath42 at frequencies below the lower cutoff frequency .
this slope is characteristic of the emission from a single synchrotron electron at frequencies below its critical frequency .
two sources ( 40.68 + 550 and 44.01 + 595 ) have spectral indices steeper than @xmath43 at the lowest measured frequencies , which requires some absorption mechanism .
since the spectral properties of the compact sources are similar , it is most probable that the mechanism responsible for the low frequency turnover is interstellar , and not a property intrinsic to each individual source .
therefore we did not test for an intrinsic lower cutoff frequency
. it would be clearly desirable to obtain even lower frequency measurements , down to 150 mhz , in order to probe the spectra most fully .
rather , we conclude that free - free absorption is the most important effect .
our spectra were fit by models involving absorbing gas , located either ( 1 ) in an intervening screen located between the source and the telescope , or ( 2 ) mixed with the synchrotron - emitting particles . from the data , it is not possible to dismiss either situation on statistical grounds alone .
to better understand the nature of this absorbing gas , we need to examine other tracers of in the nucleus of m82 .
as remarked by @xcite in an analysis of their [ ] data , the snr s are found outside of regions of highest [ ] brightness , i.e. ionized gas .
consistent with this observation , no correlation is found between the positions of the snr and the radio recombination line ( rrl ) images of @xcite .
these data trace gas that is located throughout the nuclear region of the galaxy , and perhaps concentrated in a ring , as proposed by @xcite .
we will try to estimate whether there is enough of this ambient gas to cause the absorption effects we see here .
@xcite state that the density of thermal electrons is between 10 - 100 @xmath37 .
since the snr s avoid the brightest regions of rrl emission , we assume an average value of 50 @xmath37 for interstellar thermal electrons . at a temperature @xmath44 k , the snr emission at 1 ghz would have to travel a distance of over 1 kpc to be extinguished to an optical depth of 1 .
this size represents the entire extent of the radio nucleus of m82 , and would indicate that all the `` extinguished '' snr s ( i.e. the majority of them ) were located at the far side of the nuclear region .
it is more likely that the snr s are distributed at a variety of depths into the nuclear region .
we therefore need to test the optical depth of more dense clouds . for an cloud with @xmath45 @xmath37 and @xmath44
k , at 1 ghz radio photons need travel less than 0.03 pc to be extinguished to an optical depth of 1 .
the rrl data of @xcite suggest that regions with high electron densities do exist , and that they have a very low filling factor .
we support the scenario favoured by many authors , that the ism in m82 is extremely non - uniform , in which the energy from stellar winds and sne completely dominates the state of the gas .
only extremely dense clouds have survived the disruptive effects of these processes .
these clouds cause the absorption seen at the longer wavelengths in the compact sources , either because they lie along the same line of sight to the source , or the absorbing gas is associated with the source itself .
[ [ section ] ] 44.01 + 595 + + + + + + + + + spectrum fitting of this source by @xcite suggested that it is unusual in that it has a rising spectrum towards higher frequencies .
they , and other authors ( e.g. @xcite ) consider it to be an agn candidate .
our data shows the opposite - the source has a falling , power - law spectrum between @xmath0 3.6 and 1.3 cm .
the source shows the strongest absorption in this survey , being optically thick below 2.01 ghz .
the spectrum is therefore rising below 2 ghz , and this fact is probably the cause of the previous confusion concerning the shape of the spectrum . the radio continuum spectrum alone is not a good discriminator for agn candidates .
we feel that the evidence to date concerning this source raises intriguing questions as to its identity .
a detailed vlbi study will likely be necessary to determine the exact properties of this source .
[ [ section-1 ] ] 47.37 + 680
+ + + + + + + + + this source is not included in the histogram of spectral indices ( figure 2 ) .
both powerlaw and curve fits seem to work , but the curve fit predicts fainter high - frequency emission , hence fits the sense of the data better .
this source has the shape of a ring or shell in our highest resolution maps , and therefore its integrated flux was determined by adding up the flux on a pixel - by - pixel basis using the aips task tvstat .
we show an x - band map of the source in figure 4 , at a resolution of 0@xmath323 .
the source is roughly circular , with a diameter of about 0@xmath391 , or about 16 pc .
if the remnant has been expanding undecelerated , at a rate v ( in km s@xmath2 ) , then its age would be @xmath22 3100 @xmath46 years . from its large size , this remnant would be judged the oldest one in our sample . assuming that the @xmath2250 other sources are radio snr , we estimate a supernova rate of @xmath220.016 @xmath47 yr@xmath2 , or , assuming v=5000 km s@xmath2 , one supernova explosion every 62 years . this estimate is lower than those found by other authors , however , increasingly sensitive surveys have increased the number of known compact sources , and hence our value is probably a lower limit .
we have designated five sources as flat spectrum sources .
we identify source 42.2 + 590 as an region , in support of earlier evidence presented by @xcite .
we discuss each source individually , below .
[ [ section-2 ] ] 39.11 + 573 + + + + + + + + + this source has a best - fit spectral index of @xmath48 . in the 0@xmath3050 beam of @xcite ,
it was resolved into a shell - like structure .
it is almost certainly a snr .
[ [ section-3 ] ] 42.21 + 590 + + + + + + + + + source 42.21 + 590 has many other properties typical of giant regions .
it is the site of maser activity of oh ( @xcite ) and h@xmath49o molecules ( @xcite ) .
it is also located near an emission peak at @xmath4 3.3 mm ( 92 ghz ) , at which wavelength the emission is thought to be predominantly free - free ( see @xcite ) .
we compared the source position with an image of the 12.8 @xmath23 m line of [ ] ( @xcite ) given that this line is an ideal tracer of thermal gas .
the source is indeed located within a region of high [ ] brightness .
this same [ ] emission region also encompasses the bright , confused region centred near ra 09@xmath50 51@xmath51 42@xmath525 , dec 69@xmath53 54@xmath54 58@xmath6 ( for a radio- [ ] overlay , see @xcite ) . in our radio maps ,
the two regions appear to be unconnected , but they may form part of a more massive complex .
the source 42.21 + 590 is partially resolved in the 0@xmath3050 beam at 5 ghz of @xcite , where they cite a `` largest angular size '' of 0@xmath3275 based on the diameter of their 3-@xmath7 contour .
our spectral fit predicts an optically thick flux at 1 ghz of 0.877 mjy , and assuming that the region is spherical , we derive an electron temperature of 2800 k. using equation ( 2 ) of @xcite , the predicted production rate of ionizing photons is @xmath55 @xmath56 2.43@xmath810@xmath57 @xmath58 , which is the equivalent of 108 zams o8 stars ( @xcite ) .
we summarize these properties in table 7 .
[ [ section-4 ] ] 39.40 + 561 , 45.48 + 648 , 45.74 + 652 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + these three sources can not be unambiguously identified as either radio snr or regions .
they each have a flat spectral index .
their brightness temperatures fall in the range @xmath59 k. in our highest resolution maps they are marginally resolved as a brightly peaked structure with some extended emission .
there is no evidence of any ring - like morphology .
the latter two sources fall in a region of bright , clumpy background emission , and there are several nearby point sources discernible in our highest - resolution maps .
the morphology of 45.74 + 652 is peculiar in that it has a `` tail '' of emission extending @xmath22 1@xmath30 to the north - west of the source . in a separate study
, we are undertaking a detailed examination of the large - spatial - scale , diffuse emission in m82 in order to better understand the relative importance of the thermal from the non - thermal emission .
the nature of the emission of these larger complexes of emission must be understood better before we can properly examine the spectra of those sources imbedded within them .
the results of this study will be presented in a later paper .
from our study of 26 relatively bright compact radio sources in the nucleus of m82 , we can draw the following conclusions : \2 .
high magnetic field strengths and pressures are derived for the sources we identify as snr s .
the brighter ones do not appear to be in pressure equilibrium with the surrounding medium , and are probably expanding , or are strongly confined by the high pressure ism .
the source of the high nuclear pressure and halo winds are most likely the snr s .
sources that show a low - frequency turnover , do so as a result of absorption by ionized gas .
this ionized gas may be intimately associated with the source , or may be in clouds located along the line of sight to the sources .
the gas is most likely in the form of clumpy , high density clouds .
mla wishes to thank chuck shepherd for useful discussions involving statistics and fitting procedures .
we thank richard a. sramek for the use of vla data .
it is a pleasure to thank the director and staff of the nrao for their help and generous allotment of time on the vla .
the nrao is operated by associated universities inc .
, under cooperative agreement with the national science foundation .
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 .
crrrrrrr 39.11 + 573 & 9.1(2.3 ) & 7.92(0.37 ) & 4.0(1 ) & 4.59(0.15 ) & 3.77(0.10 ) & 3.78(0.30 ) & 1.48(0.27 ) 39.40 + 561 & 4.3(1.7 ) & & & 2.05(0.15 ) & 2.02(0.10 ) & 1.10(0.73 ) & 0.63(0.33 ) 39.64 + 534 & 2.6(1.3 ) & 2.23(0.78 ) & & 1.58(0.15 ) & 0.96(0.10 ) & 0.28(0.30 ) & 0.41(0.27 ) 39.77 + 569 & 2.7(1.3 ) & 1.77(0.30 ) & & 1.06(0.15 ) & 0.75(0.11 ) & 0.20(0.30 ) & 0.34(0.27 ) 40.32 + 551 & 3.9(2.2 ) & & & 0.95(0.15 ) & 0.73(0.10 ) & 0.37(0.51 ) & 0.47(0.33 ) 40.68 + 550 & 10.9(1.6 ) & 15.37(1.18 ) & 10.0(1 ) & 7.31(0.35 ) & 6.10(0.10 ) & 4.13(0.30 ) & 2.64(0.27 ) 41.31 + 596 &
@xmath1912 & 4.15(0.33 ) & 3.0(1 ) & 3.45(0.15 ) & 2.81(0.10 ) & 1.88(0.34 ) & 1.29(0.27 ) 42.21 + 590 & @xmath1912 & & 2.5(1 ) & 4.07(0.50 ) & 4.31(0.29 ) & 4.45(0.30 ) & 4.12(0.27 ) 42.53 + 619 & 11.1(2.4 ) & 13.14(1.04 ) & & 1.64(0.15 ) & 1.52(0.50 ) & 0.25(0.30 ) & @xmath190.27 42.67 + 556 & 2.7(1.6 ) & & & 1.67(0.15 ) & 1.23(0.10 ) & 0.81(0.30 ) & 0.53(0.35 ) 43.19 + 583 & 7.9(2.3 ) & 11.65(0.58 ) & 8.5(1 ) & 5.15(0.17 ) & 3.82(0.15 ) & 2.44(0.30 ) & 1.77(0.27 ) 43.31 + 591 & 37.3(4.5 ) & 24.69(0.58 ) & 19.5(1 ) & 10.32(0.23 ) & 7.95(0.10 ) & 5.49(0.30 ) & 4.09(0.40 ) 44.01 + 595 & @xmath1912 & 8.88(0.93 ) & 9.5(1 ) & 21.79(0.89 ) & 20.18(0.10 ) & 15.79(0.30 ) & 12.48(0.27 ) 44.28 + 592 & @xmath1912 & 6.98(0.98 ) & 3.0(1 ) & 2.39(0.15 ) & 2.19(0.10 ) & 1.20(0.30 ) & 0.60(0.27 ) 44.52 + 581 & 1.6(1.5 ) & 5.15(0.30 ) & 2.5(1 ) & 2.63(0.15 ) & 1.99(0.10 ) & 1.90(0.44 ) & 0.70(0.27 ) 45.18 + 612 & 18.8(2.1 ) & 20.92(0.32 ) & 12.5(1 ) & 7.91(0.34 ) & 5.72(0.10 ) & 3.85(0.30 ) & 2.53(0.27 ) 45.25 + 651 & @xmath1912 & & 5.0(1 ) & 1.79(0.15 ) & 1.58(0.13 ) & 0.32(0.32 ) & @xmath190.27 45.43 + 673 & 7.7(0.8 ) & & 2.0(1 ) & 2.01(0.15 ) & 1.39(0.10 ) & 0.83(0.30 ) & 0.52(0.27 ) 45.48 + 648 & 2.3(1.3 ) & & & 1.32(0.15 ) & 1.16(0.10 ) & 1.46(0.41 ) & @xmath190.27 45.74 + 652 & 8.0(1.9 ) & & 2.5(1 ) & 3.40(0.28 ) & 3.48(0.48 ) & 3.23(0.50 ) & 1.33(0.58 ) 45.86 + 640 & 3.7(1.9 ) & 3.03(0.33 ) & & 1.90(0.15 ) & 1.31(0.10 ) & 0.85(0.35 ) & 0.51(0.38 ) 46.52 + 638 & 4.4(1.8 ) & 7.18(0.30 ) & 3.5(1 ) & 2.74(0.19 ) & 2.08(0.10 ) & 1.53(0.30 ) & 1.58(0.27 ) 46.56 + 738 & 3.7(0.5 ) & 2.65(0.33 ) & & 1.00(0.15 ) & 0.72(0.10 ) & 0.59(0.30 ) & 0.32(0.27 ) 46.70 + 670 & 3.9(1.5 ) & 4.17(0.45 ) & 3.0(1 ) & 1.97(0.15 ) & 1.61(0.10 ) & 1.00(0.30 ) & 1.01(0.27 ) 47.37 + 680 & 3.8(0.9 ) & 3.01(0.33 ) & & 0.91(0.15 ) & 1.09(0.20 ) & & cccccccc 39.11 + 573 & 8.62 & 0.0 & -0.38 ( 0.01 ) & 0.0 & 0.140 & 2.35 & 17.139.40 + 561 & 3.05 & 0.0 & -0.21 ( 0.01 ) & 0.0 & 0.200 & 1.37 & 5.85 39.77 + 569 & 2.13 & 0.0 & -0.49 ( 0.01 ) & 0.0 & 0.200 & 1.14 & 4.06 40.32 + 551 & 2.29 & 0.0 & -0.55 ( 0.01 ) & 0.0 & 0.200 & 1.17 & 4.25 44.28 + 592 & 6.65 & 0.0 & -0.56 ( 0.01 ) & 0.0 & 0.140 & 2.16 & 14.445.25 + 651 & 5.42 & 0.0 & -0.62 ( 0.01 ) & 0.0 & 0.220 & 1.40 & 6.05 45.43 + 673 & 4.69 & 0.0 & -0.57 ( 0.01 ) & 0.0 & 0.145 & 1.90 & 11.245.48 + 648 & 1.67 & 0.0 & -0.15 ( 0.01 ) & 0.0 & 0.200 & 1.20 & 4.44 45.74 + 652 & 5.14 & 0.0 & -0.23 ( 0.01 ) & 0.0 & 0.200 & 1.58 & 7.72 47.37 + 680 & 3.16 & 0.0 & -0.57 ( 0.01 ) & 0.0 & 0.200 & 1.29 & 5.13 crlrrcrr 39.64 + 534 & 4.42 & 1.34e-3 & -0.71 ( 0.01 ) & 0.45 ( 0.01 ) & 0.200 & 1.47 & 6.72 40.68 + 550 & 17.89 & 5.04e-4 & -0.52 ( 0.01 ) & 0.40 ( 0.01 ) & 0.240 & 1.80 & 10.041.31 + 596 & 8.59 & 4.49e-2 & -0.54 ( 0.01 ) & 1.11 ( 0.01 ) & 0.140 & 2.32 & 16.641.96 + 574 & 122.76 & 1.24e-4 & -0.72 ( 0.01 ) & 0.34 ( 0.01 ) & 0.021 & 26.4 & 2160.42.21 + 590 & 0.00 & 5.20 & -0.10 ( 0.01 ) & 2.40 ( 0.01 ) & 0.275 & & 42.53 + 619 & 30.92 & 2.51e-2 & -1.84 ( 0.01 ) & 0.65 ( 0.01 ) & 0.110 & 11.7 & 422.42.67 + 556 & 4.44 & 1.12e-3 & -0.61 ( 0.01 ) & 0.42 ( 0.01 ) & 0.195 & 1.46 & 6.61 43.19 + 583 & 15.31 & 8.68e-4 & -0.67 ( 0.01 ) & 0.45 ( 0.01 ) & 0.075 & 4.79 & 71.343.31 + 591 & 30.31 & 8.89e-5 & -0.64 ( 0.01 ) & 0.24 ( 0.01 ) & 0.195 & 2.54 & 20.144.01 + 595 & 61.99 & 2.66e-1 & -0.51 ( 0.01 ) & 2.01 ( 0.01 ) & 0.050 & 9.83 & 300.44.52 + 581 & 7.21 & 3.83e-3 & -0.61 ( 0.01 ) & 0.56 ( 0.01 ) & 0.240 & 1.40 & 6.06 45.18 + 612 & 24.08 & 4.19e-4 & -0.68 ( 0.01 ) & 0.38 ( 0.01 ) & 0.225 & 2.13 & 14.145.86 + 640 & 4.10 & 6.52e-4 & -0.53 ( 0.01 ) & 0.34 ( 0.01 ) & 0.070 & 3.39 & 35.746.52 + 638 & 9.71 & 9.74e-4 & -0.73 ( 0.01 ) & 0.46 ( 0.01 ) & 0.250 & 1.53 & 7.30 46.56 + 738 & 3.66 & 4.57e-4 & -0.78 ( 0.01 ) & 0.34 ( 0.01 ) & 0.200 & 1.44 & 6.43 46.70 + 670 & 5.22 & 4.15e-4 & -0.57 ( 0.01 ) & 0.36 ( 0.01 ) & 0.220 & 1.37 & 5.81 47.37 + 680 & 4.14 & 4.58e-3 & -0.83 ( 0.01 ) & 0.37 ( 0.01 ) & 0.200 & 1.53 & 7.28 crcrrrcrr 40.68 + 550 & 44.08 & 19.01 & 2.33e-5 & -0.55 ( 0.01 ) & 0.67 ( 0.01 ) & 0.240 & 1.83 & 10.441.31 + 596 & 3.05 & 9.07 & 3.35e-3 & -0.56 ( 0.01 ) & 1.68 ( 0.01 ) & 0.140 & 2.36 & 17.241.96 + 574 & 463.35 & 127.04 & 1.61e-1 & -0.74 ( 0.01 ) & 0.54 ( 0.01 ) & 0.021 & 26.8 & 2240.42.53 + 619 & 16.03 & 42.45 & 2.44e-3 & -1.99 ( 0.01 ) & 1.59 ( 0.01 ) & 0.110 & 15.1 & 706.42.67 + 556 & 10.66 & 4.60 & 8.02e-3 & -0.63 ( 0.01 ) & 0.67 ( 0.01 ) & 0.195 & 1.48 & 6.81 43.19 + 583 & 30.17 & 15.58 & 3.95e-1 & -0.72 ( 0.01 ) & 0.73 ( 0.01 ) & 0.075 & 4.91 & 74.743.31 + 591 & 156.73 & 29.30 & 3.94 & -0.85 ( 0.01 ) & 0.45 ( 0.01 ) & 0.195 & 2.77 & 23.844.01 + 595 & 4.48 & 72.71 & 5.32e-2 & -0.56 ( 0.01 ) & 3.77 ( 0.01 ) & 0.050 & 10.3 & 331.44.52 + 581 & 8.63 & 7.58 & 2.10e-3 & -0.64 ( 0.01 ) & 0.94 ( 0.01 ) & 0.240 & 1.43 & 6.38 45.18 + 612 & 77.12 & 24.57 & 2.37e-3 & -0.69 ( 0.01 ) & 0.58 ( 0.01 ) & 0.225 & 2.15 & 14.445.86 + 640 & 18.60 & 4.16 & 3.36e-3 & -0.54 ( 0.01 ) & 0.49 ( 0.01 ) & 0.070 & 3.41 & 36.146.52 + 638 & 9.07 & 13.53 & 1.84 & -1.50 ( 0.01 ) & 1.21 ( 0.01 ) & 0.250 & 3.17 & 31.146.56 + 738 & 10.08 & 3.69 & 3.93e-1 & -1.04 ( 0.01 ) & 0.62 ( 0.01 ) & 0.200 & 1.72 & 9.16 46.70 + 670 & 11.33 & 4.89 & 1.08 & -0.90 ( 0.01 ) & 0.67 ( 0.01 ) & 0.220 & 1.55 & 7.43 ll source name & 42.21 + 590 diameter & 0.275 optically thin flux at 10 ghz ( model fit ) & 4.398 mjy optical depth at 1 ghz ( model fit ) & 6.287 optically thick flux at 1 ghz ( model fit ) & 0.901 mjy frequency where @xmath60 = 1 ( model fit ) & 2.40 ghz brightness temperature ( 1 ghz ) & 2.10@xmath810@xmath40 k thermal luminosity & 6.932 @xmath810@xmath61 w hz@xmath2production rate of lyman continuum photons & 2.43@xmath810@xmath57 s@xmath2 emission measure at 1 ghz & 5.00@xmath810@xmath62 @xmath63 pcequivalent number of o8 stars & 108
|
we have determined detailed radio spectra for 26 compact sources in the starburst nucleus of m82 , between @xmath0 74 and 1.3 cm .
seventeen show low - frequency turnovers .
one other has a thermal emission spectrum , and we identify it as an region .
the low frequency turnovers are due to absorption by interstellar thermal gas in m82 .
new information on the agn candidate , 44.01 + 595 , shows it to have a non - thermal falling powerlaw spectrum at the highest frequencies , and that it is strongly absorbed below 2 ghz .
we derive large magnetic fields in the supernova remnants , of order 1 - 2 @xmath1 milligauss , hence large pressures in the sources suggest that the brightest ones are either expanding or are strongly confined by a dense interstellar medium . from the largest source in our sample ,
we derive a supernova rate of 0.016 yr@xmath2 .
|
the phase diagram of quantum chromodynamics ( qcd ) is investigated in large - scale lattice gauge theory simulations @xcite and heavy - ion collision experiments at cern - sps and rhic brookhaven @xcite , where the approximately baryon - free region at finite temperatures is accessible and consensus about the critical temperature for the occurence of a strongly correlated quark - gluon plasma phase ( sqgp ) is developing .
the region of low temperatures and high baryon densities , however , which is interesting for the astrophysics of compact stars , is not accessible to lattice qcd studies yet and heavy - ion collision experiments such as the cbm experiment at fair darmstadt are still in preparation @xcite .
the most stringent of the presently available constraints on the eos of superdense hadronic matter from compact stars and heavy - ion collisions have recently been discussed in ref .
@xcite and may form the basis for future systematic investigations of the compatibility of dense quark matter models with those phenomenological constraints . therefore , the question arises for appropriate models describing the nonperturbative properties of strongly interacting matter such as dynamical chiral symmetry breaking and hadronic bound state formation in the vacuum and at finite temperatures and densities .
the nambu jona - lasinio ( njl ) model has proven very useful for providing results to this question within a simple , but microscopic formulation , mostly on the mean - field level , see @xcite .
the state of the art phase diagrams of neutral quark matter for compact star applications have recently been obtained in @xcite where also references to other approaches can be found .
one of the shortcomings of the njl model is the absence of confinement , the other is its nonrenormalizability .
it is customary to speak of the njl model in its form with a cut - off regularization , where physical observables can be defined and calculated .
the cutoff in momentum space , however , defines a range of the interaction and makes the njl model nonlocal .
it has been suggested that the cutoff - regularized njl model can be considered as a limiting case of a more general formulation of nonlocal chiral quark models using separable interactions @xcite . in this form
one can even make contact with the dyson - schwinger equation approach to qcd by defining a separable representation of the effective gluon propagator @xcite , or to the instanton liquid model , see @xcite .
we have made extensive use of the parametrization given in @xcite for studies of quark matter phases in compact stars @xcite where the role of the smoothness of the momentum dependence for the quark - hadron phase transition and compact star structure has been explored .
these investigations have been also used in simulations of hybrid star cooling @xcite , which can be selective for the choice of the quation of state ( eos ) of quark matter by comparing to observational data feor surface temperature and age of compact stars . as a result of these studies , color superconducting phases with small gaps of the order of 10 kev - 1 mev
appear to be favorable for the cooling phenomenology .
a prominent candidate , the color - spin - locking ( csl ) phase , has been investigated more in detail within the njl model with satisfactory results @xcite .
however , its generalization to formfactors with a smooth momentum dependence revealed a severe sensitivity resulting in variations of the csl gaps over four orders of magnitude @xcite .
unfortunately , with the njl parametrization given in @xcite it was not possible to reproduce results with njl parametrizations given in @xcite and used , e.g. , in refs .
@xcite . therefore , in the present work a new parametrization of the model presented in ref .
@xcite is performed with a special emphasis on reproducing njl parametrizations given in @xcite in the limiting case of a sharp cutoff formfactor .
we also take into account the strangeness degree of freedom and consider lorentzian - type formfactor models where the form of the momentum dependence for the quark - quark interaction can be varied parametrically thus being most suitable for a quantitative analysis the phase diagram and high - density eos under the above mentioned constraints from compact star and heavy - ion collision phenomenology .
we consider a nonlocal chiral quark model with separable quark - antiquark interaction in the color singlet scalar / pseudoscalar isovector channel @xcite where the formfactors are given in the instantaneous approximation , in the same way as it was suggested in @xcite . the lagrangian density of the quark model is given by ( @xmath2 ) @xmath3,\ ] ] where indices occuring twice are to be summed over and the formfactor @xmath4 for the nonlocal current - current coupling has been introduced .
here @xmath5 and @xmath6 are the current quark masses of the light and strange flavors , respectively , @xmath7 are the gell - mann matrices of the @xmath8 flavor group and @xmath9 , @xmath10 are dirac matrices . the nonlocality of the current - current interaction in the quark - antiquark ( @xmath11 ) channel is implemented in the separable approximation via the same formfactor functions for all colors and flavors . in our calculations
we use the gaussian ( g ) , lorentzian ( l ) , woods - saxon ( ws ) and cutoff ( njl ) formfactors in momentum space defined as ( see ref .
@xcite ) @xmath12^{-1 } , \nonumber \\
g_{{\rm ws}}(p)&= & [ 1+\exp(- \alpha)]/\{1+\exp[\alpha~ ( p^2/\lambda_{{\rm ws}}^2 - 1]\},\nonumber \\ g_{{\rm njl}}(p ) & = & \theta ( 1-p/\lambda _ { { \rm njl}})~.\nonumber\end{aligned}\ ] ] the formfactors can be introduced in a manifestly covariant way ( see @xcite ) , but besides technical complications at finite @xmath13 and @xmath14 , where matsubara summations have to be performed numerically , it is not a priori obvious that such a formulation shall be superior to an instantaneous approximation ( 3d ) which could be justified as a separable representation of a coulomb - gauge potential model @xcite .
typically , three - flavor njl type models use a t hooft determinant interaction that induces a u@xmath15(1 ) symmetry breaking in the pseudoscalar isoscalar meson sector , which can be adjusted such that the @xmath16-@xmath17 mass difference is described . in the present approach
this term is neglected using the motivation given in @xcite , so that the flavor sectors decouple in the mean - field approximation .
the dynamical quark mass functions are then given by @xmath18 , where the chiral gaps fulfill the gap equations @xmath19 corresponding to minima of the thermodynamic potential with respect to variations of the order parameters @xmath20 , the quark dispersion relations are @xmath21 .
the basic set of equations should be chosen to fix the parameters included in the model , which are the current masses , coupling constant and cutoff parameter ( @xmath22 and @xmath1 ) . in order to do that we use the properties of bound states of quarks in the vacuum given by the pion decay constant @xmath23 mev , the masses of the pion @xmath24 mev and the kaon @xmath25 mev and either the constituent quark mass @xmath26 or the chiral condensate of light quarks , defined as @xmath27 with a phenomenological value from qcd sum rules @xcite of @xmath28 mev @xmath29 mev .
the chiral condensate generally is not properly defined in the case of nonlocal interactions .
the subtraction of the @xmath30 term has been included to make the integral convergent .
the pion decay constant can be expressed in the form @xmath31where the pion wave function renormalization factor @xmath32 is @xmath33 the masses of pion and kaon are obtained from a direct generalization of the well - known njl model @xcite by introducing formfactors with the momentum space integration and replacing constituent quark masses by the momentum dependent mass functions @xmath34
@xmath35 ^{1/2 } , \\
m_{k } & = & \left [ \left ( \frac{1}{2g_{s}}-(i_{u}^{(1)}+i_{s}^{(1)})\right ) /i_{us}^{(2)}\right ] ^{1/2},\end{aligned}\ ] ] in these mass formulae , the following abbreviations for integrals have been used @xmath36}.\end{aligned}\ ] ] we can use these notations to give an estimate of the validity of low - energy theorems for this nonlocal generalization of the njl model . to this end
we rewrite eq .
( [ fpi ] ) for @xmath37 and @xmath38 as @xmath39 where the mean values of a distribution @xmath40 are defined using the integral @xmath41 as an operator : @xmath42 $ ] . to leading order in an expansion at the chiral limit ( @xmath43 , @xmath44 ) one obtains the goldberger - treiman relation @xmath45 rewriting the gap equation ( [ disprel ] ) for the light flavor as @xmath46=- m_0 \langle
u\bar{u } \rangle_0 \cdot 4 g_s\ ] ] and the pion mass formula ( [ mpi ] ) as @xmath47 we obtain by combining ( [ mpi2 ] ) with ( [ gap ] ) , ( [ gt ] ) and ( [ fpi-2 ] ) in leading order the gell - mann oakes renner relation ( gmor ) @xmath48 as an indicator of the validity of this low - energy theorem we will show the gmor value for the light current quark mass @xmath49 together with the result of the parametrization of @xmath30 .
since we have no t hooft term , there is no mixing of flavor sectors , and one can consider the light quark sector independent of the strange one . the equation for the kaon mass fixes the strange quark s current mass @xmath6 , whereby a self - consistent solution of the strange quark gap equation is implied .
in the present parametrization scheme the gap equation plays a special role .
although the gap is not an observable quantity , we will use it as an phenomenological input instead of the condensate , which in some cases does not fulfill the phenomenological constraints .
moreover , for each formfactor model there is some minimal value of @xmath50 for which the condensate has a minimum : for the gaussian model it is 7.376 , for the lorentzian model with @xmath51 it is 3.795 , and for @xmath52 it is 2.825 . for the njl model
this minimal value is 2.588 .
the corresponding values of the condensate are given in table [ lorsysm ] .
these values for the finite current masses are shifted to the left as it is shown in fig .
[ patalogy ] and for them the parameter sets are fixed ( see table [ lorsysm ] ) .
when the condensate is chosen there are two possible values of @xmath50 ( the lower and higher branches ) for which one can fix the parameters of the model .
we show that the constraint on the condensate from qcd sum rules @xcite with an upper limit at @xmath53 mev can be fulfilled only for values of @xmath54 exceeding 3 - 5 for both lorentzian and woods- saxon formfactor models . for the particular choice of @xmath55 mev and 260 mev we fixed the parameters for both branches of solutions ( see tables [ ll260 ] -[njl ] ) . in order to obtain the parameter sets we choose three values for the non - observable value of the constituent quark mass @xmath56 mev , 335 mev , 367.5 mev , 380 mev , and 400 mev .
the values are taken such that the mass @xmath57 is larger than the mass of the nucleon as a bound state of three quarks .
the results of the parametrizations are given in the tables [ l330]-[l400 ] . in fig .
[ patalogy ] the dependence of the chiral condensate is shown as a function of @xmath58 for different formfactors in the chiral limit band for an appropriate choice of the current mass @xmath59 .
it is shown that the minimal possible value of the condensate varies from one formfactor model to another and only in the njl model the appropriate values of condensate in the range of qcd sum rule values @xmath60 mev can be reached . the figs .
[ ffcomp ] and [ ffpcomp ] show the gap function and the diagonal elements of the separable interaction for different formfactors in order to demonstrate the systematics of the changes related to the degree of the softening given by the parameter @xmath54 in lorentzian functions .
we have presented parametrizations of nonlocal chiral quark models with instantaneous , separable interactions defined by momentum dependent formfactors which interpolate between the soft gaussian type and the hard cut off ( njl ) in tabulated form .
the introduction of a lorentzian and/or woods - saxon - type function with an additional parameter allowed a systematic investigation of the njl model limit , where existing parametrizations could be recovered .
we have shown that the instantaneous nonlocal models have an essential problem for the softest formfactors , where it is impossible to obtain acceptable values for the chiral condensate . however , for the astrophysical applications this problem could be considered as of minor importance relative to the insights which a systematic variation of the interaction model offers for the better understanding of mechanisms governing the quark matter eos on a microscopic level .
the present approach to nonlocal chiral quark models can be applied subsequently for systematic studies of constraints on the eos of superdense matter coming from the phenomenology of heavy ion collisions and compact stars .
i thank david blaschke , norberto scoccola and yuri kalinovsky for the initiation of this work and their constructive discussions .
i am grateful to a. dorokhov , o. teryaev and v. yudichev for their interest in this work and support during my visit at the jinr dubna .
the research was supported in part by dfg under grant no .
436 arm 17/4/05 and by the daad partnership program between the universities of rostock and yerevan .
t. klhn _ et al .
_ , arxiv : nucl - th/0602038 . m. buballa , phys .
* 407 * ( 2005 ) 205 .
s. b. rster , v. werth , m. buballa , i. a. shovkovy and d. h. rischke , phys .
d * 72 * ( 2005 ) 034004 .
d. blaschke , s. fredriksson , h. grigorian , a. m. ztas and f. sandin , phys . rev .
d * 72 * ( 2005 ) 065020 .
d. n. aguilera , d. blaschke and h. grigorian , nucl .
phys . a * 757 * ( 2005 ) 527 .
h. abuki and t. kunihiro , nucl .
a * 768 * ( 2006 ) 118 .
s. m. schmidt , d. blaschke and y. l. kalinovsky , phys .
c * 50 * ( 1994 ) 435 .
d. blaschke , g. burau , y. l. kalinovsky , p. maris and p. c. tandy , int .
j. mod .
phys . a * 16 * ( 2001 ) 2267 .
d. gomez dumm , d. b. blaschke , a. g. grunfeld and n. n. scoccola , arxiv : hep - ph/0512218 .
d. blaschke , s. fredriksson , h. grigorian and a. m. ztas , nucl .
a * 736 * ( 2004 ) 203 .
h. grigorian , d. blaschke and d. n. aguilera , phys .
c * 69 * ( 2004 ) 065802 .
h. grigorian , d. blaschke and d. voskresensky , phys .
c * 71 * ( 2005 ) 045801 . s. popov , h. grigorian and d. blaschke , arxiv : nucl - th/0512098 .
d. n. aguilera , d. blaschke , m. buballa and v. l. yudichev , phys .
d * 72 * ( 2005 ) 034008 [ arxiv : hep - ph/0503288 ] .
d. n. aguilera and d. b. blaschke , arxiv : hep - ph/0512001 . c. gocke , d. blaschke , a. khalatyan and h. grigorian , arxiv : hep - ph/0104183 .
.parameter sets for the chiral quark model with gaussian , woods - saxon and lorentzian formfactors for different values of the @xmath61 parameter ( including cutoff ) , when @xmath56 mev . [ cols="^,^,^,^,^,^,^,^,^ " , ]
|
in these notes we describe the basic formulation of the parametrization scheme for the instantaneous nonlocal chiral quark model in the three flavor case . we choose to discuss the gaussian , lorentzian - type , woods - saxon and sharp cutoff ( njl ) functional forms of the momentum dependence for the formfactor of the separable interaction .
the four parameters : light and strange quark mass , coupling strength ( @xmath0 ) and range of the interaction ( @xmath1 ) have been fixed by the same phenomenological inputs : pion and kaon mass , pion decay constant and light quark mass in vacuum .
the wood - saxon and lorentzian - type formfactor is suitable for an interpolation between sharp cutoff and soft momentum dependence .
results are tabulated for applications in models of hadron structure and quark matter at finite temperatures and chemical potentials where separable models have proven successful .
|
spin - triplet superconductivity , which is arousing much interests in recent years , is fascinating in a number of ways .
theoretically , a most intriguing question is the following : if we consider the electron mechanism of superconductivity in the most frequently adopted hubbard model with an on - site electron - electron repulsion , we can show that triplet superconductivity is very difficult to realize for the simple reason that the pairing interaction mediated by spin fluctuations is only 1/3 in the triplet channel than in the singlet channel@xcite .
so any theory attempting to explain triplet pairing has to overcome this question .
experimentally , the discovery of superconductivity by maeno and coworkers in the layered perovskite ruthenium oxide sr@xmath8ruo@xmath9 have kicked off renewed interests . suggestions for a triplet pairing in this material came from nmr knight shift@xcite and polarized neutron scattering@xcite . a broken time reversal symmetry is further observed with @xmath10sr@xcite and small - angle neutron scattering@xcite . as for the pairing symmetry , nmr and nqr relaxation rates have shown an absence of the hebel - slichter peak@xcite , which suggests line node(s ) in the gap function , which was supported by a specific heat measurement@xcite . while the angular dependence of thermal conductivity indicates the presence of horizontal line - nodes@xcite , a recent field - orientation dependence of the specific heat shows that the gap in the active @xmath7 band has minima along [ 100 ] directions with the passive @xmath11 and @xmath12 bands having gap minima along [ 110 ] directions@xcite .
a theoretical work by zhitomirsky and rice@xcite has proposed horizontal line nodes with a good fit to the specific heat measurement@xcite . as for the mechanism that stabilizes triplet pairing , kuwabara and ogata@xcite , and independently sato and kohmoto@xcite , have suggested that triplet @xmath13-wave pairing can be induced by an anisotropy in the spin fluctuation .
kuwabara and ogata@xcite have identified the most suitable gap function to be @xmath14 that breaks the time - reversal symmetry , which was also proposed by miyake and narikiyo@xcite .
the chiral @xmath15 ( @xmath16 : @xmath13-wave like ) has also been obtained with a third - order perturbation theory by nomura and yamada@xcite and more recently by yanase and ogata@xcite , and with the fluctuation exchange approximation ( flex ) for quasi - one - dimensional @xmath11 and @xmath12 bands by kuroki _
et al._@xcite . however , the validity of perturbation results truncated at finite orders has to be checked by non - perturbative methods . indeed , a recent quantum monte carlo ( qmc ) result by kuroki _ et al . _ have shown that the singlet @xmath17-pairing dominates over the triplet for the @xmath7 band.@xcite this situation has led arita _
_ to show , with the dynamical cluster approximation ( dca ) , that @xmath18 pairing actually dominates over @xmath13-wave pairings as far as the on - site hubbard model is concerned , while if we go over to the extended hubbard model with off - site repulsion the triplet superconductivity can be favored but not dominant in extended hubbard model employing nearest neighbor coulomb repulsion for @xmath7 band in sr@xmath8ruo@xmath9.@xcite generally , there is a flex result by arita et al@xcite who show that triplet superconductivity is much weaker than singlet pairs for the one - band , on - site hubbard model , which agrees with a phenomenology@xcite , while if charge fluctuations are enhanced by off - site repulsions triplet pairs have a chance to dominate .
we have already shown triplet superconductivity can be dominant near the charge density wave ( cdw ) phase on square lattice with the flex in the extended hubbard model.@xcite so the purpose of this paper is to examine how triplet superconductivity can become dominant in the extended hubbard model with the flex in general , and for @xmath7 band of sr@xmath8ruo@xmath9 in particular . the extended hubbard model has been studied , primarily for specific charge densities @xmath19 , e.g. , half - filling or quarter filling , by means of the quantum monte carlo method@xcite , the weak coupling theory@xcite , the mean - field approximation@xcite , the second - order perturbation@xcite , the random phase approximation@xcite , the flex approximation@xcite , the slave - boson technique@xcite , the bosonization and the renormalization group@xcite . here
we adopt the flex developed by bickers _
et al._@xcite , which is a renormalized perturbation method to study pairing instabilities when exchange of spin and charge fluctuations are considered as dominant diagrams .
this approximation is useful to explore the tendencies of dominant pairing relatively when the parameters are systematically varied .
however , including the off - site coulomb repulsion @xmath1 susceptibilities and effective interactions become @xmath20 matrices for the lattice coordination number @xmath21 ( @xmath22 for the square lattice ) , which demand extremely computer resources .
we show as far as the present finite - temperature result is concerned that ( i ) triplet @xmath4 pairing appears in between singlet @xmath23 and the cdw phase for @xmath24 , and ( ii ) the pairing symmetry changes as @xmath18 @xmath25 @xmath4 @xmath25 @xmath23 when the shape of the fermi surface is varied by the second - neighbor hopping @xmath2 .
physically , all the results can be well explained by the structure and peak value of spin and charge susceptibilities .
let us start with the extended hubbard hamiltonian , @xmath26 in the standard notation on a tetragonal lattice depicted in fig.[lattice ] , where nn ( nnn ) denotes nearest neighbor ( next - nearest neighbor ) sites .
for the square lattice the unit of energy is taken to be the nearest - neighbor @xmath27 , and lattice constant @xmath28 . to determine the dominant gap function
, we solve liashberg s equation with the flex approximation , @xmath29 where @xmath30 is the gap function , @xmath31 green s function , and @xmath32 the pairing interaction with @xmath33 .
the eigenvalue @xmath34 , a measure of the pairing , becomes unity at @xmath35 . for the calculation
we take an @xmath36 lattice , the temperature @xmath37 , and the matsubara frequency for fermions @xmath38 with @xmath39 .
esirgen _ et al._@xcite have extended the flex method to general lattice hamiltonians including the extended hubbard model . following them we introduce the pairing interaction , @xmath40(k - k';\delta{\mbox{\boldmath$r$}};\delta{\mbox{\boldmath$r$}}')e^{i({\mbox{\boldmath$k$}}\cdot\delta{\mbox{\boldmath$r$}}+{\mbox{\boldmath$k'$}}\cdot\delta{\mbox{\boldmath$r$}}')}\nonumber\\ -&\frac{1}{2}&\left[v_{\rm ch}\chi_{\rm ch}v_{\rm ch}\right](k - k';\delta{\mbox{\boldmath$r$}};\delta{\mbox{\boldmath$r$}}')e^{i({\mbox{\boldmath$k$}}\cdot\delta{\mbox{\boldmath$r$}}+{\mbox{\boldmath$k'$}}\cdot\delta{\mbox{\boldmath$r$ } } ' ) } \nonumber\\ + & \frac{1}{2}&v_{\rm s}(0;\delta{\mbox{\boldmath$r$}};\delta{\mbox{\boldmath$r$}}')e^{i({\mbox{\boldmath$k$}}\cdot\delta{\mbox{\boldmath$r$}}'-{\mbox{\boldmath$k'$}}\cdot\delta{\mbox{\boldmath$r$ } } ) } \left.\right\ } , \label{s}\end{aligned}\ ] ] for singlet pairing , and @xmath41(k - k';\delta{\mbox{\boldmath$r$}};\delta{\mbox{\boldmath$r$}}')e^{i({\mbox{\boldmath$k$}}\cdot\delta{\mbox{\boldmath$r$}}+{\mbox{\boldmath$k'$}}\cdot\delta{\mbox{\boldmath$r$}}')}\nonumber\\ -&\frac{1}{2}&\left[v_{\rm ch}\chi_{\rm ch}v_{\rm ch}\right](k - k';\delta{\mbox{\boldmath$r$}};\delta{\mbox{\boldmath$r$}}')e^{i({\mbox{\boldmath$k$}}\cdot\delta{\mbox{\boldmath$r$}}+{\mbox{\boldmath$k'$}}\cdot\delta{\mbox{\boldmath$r$}}')}\nonumber\\ + & \frac{1}{2}&v_{\rm t}(0;\delta{\mbox{\boldmath$r$}};\delta{\mbox{\boldmath$r$}}')e^{i({\mbox{\boldmath$k$}}\cdot\delta{\mbox{\boldmath$r$}}'-{\mbox{\boldmath$k'$}}\cdot\delta{\mbox{\boldmath$r$ } } ) } \left.\right\ } \label{t}\end{aligned}\ ] ] for triplet pairing .
here @xmath42 is null or nearest - neighbor vectors , @xmath43 are the spin and charge susceptibilities , respectively , where @xmath44 is the irreducible susceptibility , @xmath45 and @xmath46 is the coupling between density ( magnetic ) fluctuations , @xmath47 , & \delta{\mbox{\boldmath$r$}}=\delta{\mbox{\boldmath$r$}}'={\mbox{\boldmath$0$}},\\ -v_x , & \delta{\mbox{\boldmath$r$}}=\delta{\mbox{\boldmath$r$}}'=\pm\hat{{\mbox{\boldmath$x$}}}\\ -v_y , & \delta{\mbox{\boldmath$r$}}=\delta{\mbox{\boldmath$r$}}'=\pm\hat{{\mbox{\boldmath$y$}}}\\ \end{array } \right.\\ & v_{\rm sp}&(q;\delta{\mbox{\boldmath$r$}};\delta{\mbox{\boldmath$r$}}')= \left\ { \begin{array}{ll } -u , & \delta{\mbox{\boldmath$r$}}=\delta{\mbox{\boldmath$r$}}'={\mbox{\boldmath$0$}},\\ -v_x , & \delta{\mbox{\boldmath$r$}}=\delta{\mbox{\boldmath$r$}}'=\pm\hat{{\mbox{\boldmath$x$}}}\\ -v_y , & \delta{\mbox{\boldmath$r$}}=\delta{\mbox{\boldmath$r$}}'=\pm\hat{{\mbox{\boldmath$y$}}},\\ \end{array } \right.\end{aligned}\ ] ] where @xmath48 with @xmath49 being the matsubara frequencies for bosons , and @xmath50 here .
we have found that the @xmath51 dependence of @xmath52 and @xmath53 does not significantly affect @xmath54 and @xmath55 .
accordingly the peak position of @xmath56 is almost the same as that for @xmath57 term in the expression for @xmath32 .
@xmath58 , @xmath59 , appearing in the last lines in eqs.([s],[t ] ) respectively , are constant terms involving @xmath0 and @xmath1 , @xmath60 when the off - site interaction @xmath1 is introduced all the vertices ( @xmath52 , @xmath53 , @xmath61 , @xmath62 ) as well as the susceptibilities become @xmath20 matrices for the lattice coordination number @xmath63 for the square lattice ) .
we consider @xmath7 band of sr@xmath8ruo@xmath9 whose band filling is @xmath64 . to represent the shape of the fermi surface obtained by arpes@xcite
, we choose parameters @xmath65 and next nearest hopping @xmath66 .
we first show the phase diagram against on - site coulomb interaction @xmath0 and the nearest neighbor coulomb interaction @xmath1 ( fig . [ phase ] ) .
there , we have assumed that the dominant superconductivity pairing is the one that has the largest eigenvalue @xmath34 in liashberg s equation calculated at @xmath37 . while the value of @xmath34 for @xmath37 is still much smaller than unity , it is difficult to extend the flex calculation to lower temperatures , given the complexity of the ( extended hubbard ) model .
so this amounts to an assumption that the order in which @xmath34 s appear for various pairing symmetries does not change for @xmath67 .
the charge density wave ( cdw ) is identified as the region in which the charge susceptibility ( which has a peak at @xmath68 ) diverges at @xmath37 . in the result a triplet superconductivity phase ( which does not exist for a single band when @xmath1 is absent ) is seen to appear just below the cdw phase for @xmath24 .
the maximum eigenvalue of charge and spin susceptibilities and detailed behavior of @xmath34 as a function of @xmath1 for fixed @xmath69 is shown in fig .
[ su - csdw ] . from the figure
we see that the triplet superconductivity becomes dominant when the charge susceptibility @xmath56 is larger than the spin susceptibility @xmath70 .
figure [ gap ] shows gap functions in @xmath71-space , which shows that the triplet ( which has the highest @xmath34 ) has a gap function @xmath72 ) , while the two singlet gap functions have @xmath23 and @xmath73.@xcite we can trace back the reason why these gap functions are favored to the structure of spin and charge susceptibilities .
figure [ peak ] shows the peak positions of spin and charge susceptibilities for @xmath74 .
the peak positions for charge are around @xmath75 , while those for spin are ferromagnetic - like around @xmath76 , @xmath77 . in liashberg s eq.([elia])-([t ] ) , we see that for singlet the coefficient of @xmath70 ( @xmath56 ) term in @xmath78 are positive ( negative ) .
so the dominant gap function must change sign across the momentum transfer that points the peak position in the spin susceptibility , while the gap function must not change sign across the momentum transfer that points the peak position in the charge susceptibility . for triplet ,
on the other hand , the coefficients of @xmath70 and @xmath56 terms in @xmath55 are both negative .
so the gap function should have the same sign across both the spin and charge peak positions .
when @xmath1 is small , the spin susceptibility is much greater than the charge susceptibility , and the spin - fluctuation mediated pairing interaction for singlet is three times larger than that for triplet .
this explains @xmath23 , which changes sign across @xmath76 , @xmath77 , becomes the dominant gap function . for large @xmath1 ,
the charge susceptibility becomes dominant , and the pairing interaction becomes the same between singlet and triplet channels , i.e. , @xmath79 and @xmath80 become similar in magnitude . whether triplet is more favored than singlet depends on the other factors such as the number of nodes on the fermi surface , which work unfavorably for pairing since the integration around the node in the right hand side of eqn.([elia ] ) can not contribute to @xmath34 .
so a larger the number of nodes is basically unfavored . as another factor
, we can find from fig .
[ su - csdw ] that the @xmath34 s for the gap function @xmath4 and @xmath73 increase more sharply than that for @xmath23 for @xmath81 , where @xmath56 increases sharply .
the reason should be that the first two have the same sign across the peak of @xmath56 around @xmath82 or @xmath83 , while the latter has slightly warped nodal lines .
so there is a region exists where the gap function has the opposite signs across the peak of @xmath56 .
returning to sr@xmath8ruo@xmath9 , our result in fig .
[ su - csdw ] indicates that if triplet can be dominant , the gap function should be @xmath4 which is degenerate with @xmath5 .
the true gap function below @xmath84 should be a complex linear combination , @xmath85 which is more stable thermodynamically and breaks the time - reversal symmetry ( fig .
[ tri ] ) .
the absolute value of the gap function has minima on the fermi surface along @xmath86 $ ] and equivalent directions as depicted in fig .
[ tri ] with open circles , which is consistent with the recent experiment of the specific heat in rotated magnetic fields.@xcite however , we also notice that the triplet region is very narrow , so identification of the pairing in sr@xmath8ruo@xmath9 would require a precise determination of the parameters in that material . in addition , if we include the effect of @xmath11 and @xmath12 band of sr@xmath8ruo@xmath9 this triplet region may expand .
let us identify the relation between the pairing symmetry and shape of the fermi surface in the present context .
for that purpose we can change the second neighbor hopping @xmath2 , which controls the warping ( and even the topology ) of the fermi surface . in fig .
[ su - t ] we show @xmath34 and the dominant pairing symmetry when @xmath2 is varied with fixed @xmath87 and @xmath74 .
figure [ fermi - gap - t ] depicts how the fermi surface obtained by @xmath88 changes with @xmath2 .
the reason why those gap functions dominate can be understood in terms of the structure and value of the spin and charge susceptibilities . while the peaks in the spin susceptibility for @xmath89 correspond to the nesting vectors for the fermi surface
, the peaks for large @xmath1 do not necessarily correspond to those .
the result shows that the peaks change from @xmath90 , @xmath91 for @xmath92 , @xmath93 for @xmath94 , to @xmath95 , @xmath96 for @xmath97 as seen in fig .
[ sdw - peak ] . on the other hand
the peak position in the charge susceptibility remains at @xmath68 for the whole range of @xmath2 .
the peak values of @xmath98 and @xmath99 are shown in fig .
[ csdw - t ] , where the peak of @xmath100 changes rapidly while @xmath101 has an almost constant peak value . from the result
we see that nearly antiferromagnetic spin fluctuations favor @xmath18 pairing for @xmath92 as in the high-@xmath84 cuprates , strong charge fluctuations favor @xmath4 for @xmath102 . for @xmath103 where the spin susceptibility again exceeds the charge susceptibility , nearly ferromagnetic spin fluctuations favor @xmath23 which has many sign changes on the fermi surface .
finally it is interesting to identify the reason why the charge fluctuation is sharply peaked around @xmath104 . to elaborate the point , we first note the van - hove singularities , which reside at @xmath105 , @xmath106 , @xmath107 and @xmath108 for the @xmath109 tight - binding model .
so the charge susceptibility with a peak around @xmath68 should be maximized when the fermi surface approaches @xmath105 and @xmath106 , since the pair hopping with the momentum transfer @xmath68 connects the van - hove singularities , where the density of state is large . in the terminology of the previous work of ours@xcite ,
the fermi surface becomes `` thick '' around the van - hove singularities .
we have actually studied the relation between the charge susceptibility and the fermi surface as shown in fig .
[ van ] to confirm that the charge susceptibility becomes maximum for @xmath110 , at which the fermi surface just touches the van - hove singularities @xmath105 , @xmath106 and the topology of the fermi surface changes .
we see in the figure that the fermi surface becomes `` thick '' around the van - hove singularities .
the value @xmath104 is close to the @xmath2 for @xmath7 band of sr@xmath8ruo@xmath9 , which may imply that triplet superconductivity induced by charge fluctuations may in fact be relevant there .
so far we have shown all the results in @xmath111-space , but it is also heuristic to represent the pairing symmetry in real space as depicted in fig .
[ real ] . from fig .
[ su - t]-[csdw - t ] we see that @xmath18 is suppressed as the @xmath68 charge fluctuation increases , while from fig .
[ su - csdw ] @xmath4 , @xmath112 and @xmath23 are favored as the charge fluctuation increases .
this should be because strong @xmath68 charge fluctuations tend to arrange electrons diagonally on the lattice , and suppresses @xmath18 pairs across the nearest neighbors , but do not disturb the pairing such as @xmath4 , @xmath112 and @xmath23 formed across more distant sites .
we have studied the pairing symmetry in the one - band extended hubbard model having the nearest neighbor coulomb repulsion for square lattice . for the parameter which represents @xmath7 band of sr@xmath8ruo@xmath9 , spin - triplet superconductivity
whose symmetry is @xmath4 appears just below the cdw phase .
so the true gap below @xmath84 is suggested to be a chiral @xmath6 which breaks the time - reversal symmetry .
it would be interesting to experimentally detect the pairing state using , for example , a theoretical prediction made in ref . .
however , the region for the spin - triplet superconductivity is very small , and the identification of the pairing in sr@xmath8ruo@xmath9 will require a precise determination of parameters , and also a study of the effects of @xmath11 and @xmath12 bands .
we have also examined the relation between the pairing symmetry and the shape of the fermi surface by varying @xmath2 .
as @xmath2 increases , the shape of the fermi surface changes drastically which changes the structure of @xmath70 and @xmath56 .
this in turn changes the pairing symmetry as @xmath113 .
in addition , @xmath56 becomes especially large when the fermi surface touches van - hove singularities at which the fermi surface changes the topology , which is achieved for @xmath110 near the parameter for @xmath7 band of sr@xmath8ruo@xmath9 .
numerical calculations were performed at the supercomputer center , issp .
@xmath114 present address : graduate school of engineering , nagoya university , chikusa , nagoya 464 - 8603 , japan .
+ @xmath115 present address : max - planck - institut fr festkoperforschung , stuttgart , germany .
r. arita , k. kuroki , and h. aoki , phys .
b * 60 * , 14585 ( 1999 ) ; j. phys .
. jpn . * 69 * , 1181 ( 2000 ) .
p. monthoux and g. lonzarich , phys .
b * 59 * , 14598 ( 1999 ) .
k. ishida , h. mukuda , y. kitaoka , k. asayama , z. q. mao , y. mori , and y. maeno , nature ( london ) * 396 * , 658 ( 1998 ) .
k. ishida , h. mukuda , y. kitaoka , z. q. mao , h. fukazawa , and y. maeno , phys .
b * 63 * , 060507(r ) ( 2001 ) . j. a. duffy , s. m. hyden , y. maeno , z. mao , j. kulda , and g. j. mcintyre , phys . rev
. lett . * 85 * , 5412 ( 2000 ) .
g. m. luke , y. fudamoto , k. m. kojima , m. i. larkin , j. merrin , b. nachumi , y. j. uemura , y. maeno , z. q. mao , y. mori , h. nakamura , and m. sigrist , nature ( london ) * 394 * , 448 ( 1998 ) .
g. m. luke , y. fudamoto , k. m. kojima , m. i. larkin , b. nachumi , y. j. uemura , j. e. sonier , y. maeno , z. q. mao , y. mori , h. nakamura , and d. agterberg , physica b * 289 * , 373 ( 2000 ) .
g. kealey , t. m. riseman , e. m. forgan , a. p. mackenzie , l. m. galvin , s. l. lee , d. mck .
paul , r. cubitt , d. f. agterberg , r. heeb , z. q. mao , and y. maeno , phys .
lett . * 84 * , 6094 ( 2000 ) .
k. ishida , y. kitaoka , k. asayama , s. ikeda , s. nishizaki , y. maeno , k. yoshida , and t. fujita , phys .
b * 56 * , r505 ( 1997 ) .
k. ishida , h. mukuda , y. kitaoka , z. q. mao , h. fukazawa , and y. maeno , phys .
lett . * 84 * , 5387 ( 2000 ) .
s. nishizaki , y. maeno , and z. q. mao , j. phys .
. jpn . * 69 * , 572 ( 2000 ) .
k. izawa , h. takahashi , h. yamaguchi , y. matsuda , m. suzuki , t. sasaki , t. fukase , y. yoshida , r. settai , and y. onuki , phys .
* 86 * , 2653 ( 2001 ) .
k. deguchi , z. q. mao , h. yaguchi , and y. maeno , phys .
* 92 * , 047002 ( 2004 ) .
m. e. zhitomirsky and t. m. rice , phys .
* 87 * , 057001 ( 2001 ) .
t. kuwabara and m. ogata , phys .
lett . * 85 * , 4586 ( 2000 ) .
m. sato and m. kohmoto , j. phys
. soc . jpn .
* 69 * , 3505 ( 2000 ) .
k. miyake and o. narikiyo , phys .
lett . * 83 * , 1423 ( 1999 ) .
t. nomura and k. yamada , j. phys .
. jpn . * 69 * , 3678 ( 2000 ) .
t. nomura and k. yamada , j. phys .
. jpn . * 71 * , 1993 ( 2002 ) .
y. yanase and m. ogata , j. phys .
* 72 * , 473 ( 2003 ) .
k. kuroki , m. ogata , r. arita , and h. aoki , phys .
b * 63 * , 060506 ( 2001 ) .
k. kuroki , y. tanaka , t. kimura , and r. arita , phys .
b * 69 * , 214511 ( 2004 ) .
r. arita , s. onari , k. kuroki , and h. aoki , phys .
lett . * 92 * , 247006 ( 2004 ) .
s. onari , r. arita , k. kuroki , and h. aoki , phys .
b * 70 * , 094523 ( 2004 ) .
y. zhang and j. callaway , phys .
b * 39 * , 9397 ( 1989 ) .
z. teanovi , a. r. bishop , and r. l. martin , solid state commun .
* 68 * , 337 ( 1988 ) .
m. murakami , j. phys .
jpn . * 69 * , 1113 ( 2000 ) .
h. seo and h. fukuyama , j. phys .
jpn . * 66 * , 1249 ( 1997 ) .
m. onozawa , y. fukumoto , a. oguchi , and y. mizuno , phys .
b * 62 * , 9648 ( 2000 ) .
d. j. scalapino , e. loh , jr . , and j. e. hirsch , phys . rev .
b * 35 * , 6694 ( 1987 ) .
g. esirgen , h.b .
schuttler , and n. e. bickers , phys .
. lett . * 82 * , 1217 ( 1999 ) .
j. merino and r. h. mckenzie , phys .
lett . * 87 * , 237002 ( 2001 ) .
k. sano and y. ono , j. phys .
solids * 63 * , 1567 ( 2002 ) .
k. kuroki , k. kusakabe , and h. aoki , phys .
b * 50 * , 575 ( 1994 ) . n. e. bickers , d. j. scalapino , and s.r .
white , phys .
* 62 * , 961 ( 1989 ) .
n. e. bickers and d. j. scalapino , ann .
* 193 * , 206 ( 1989 ) .
t. dahm and l. tewordt , phys .
b * 52 * , 1297 ( 1995 ) .
m. langer , j. schmalian , s. grabowski , and k. h. bennemann , phys .
* 75 * , 4508 ( 1995 ) .
g. esirgen and n. e. bickers , phys .
b * 55 * , 2122 ( 1997 ) . g. esirgen and n. e. bickers , phys .
b * 57 * , 5376 ( 1998 ) .
a. damascelli , _ et al .
* 85 * , 5194 ( 2000 ) . in ref.@xcite
the dominant spin singlet gap function is @xmath18 for @xmath116 , @xmath104 with @xmath89 , where the spin structure is antiferromagnetic ( peaked at @xmath117 ) , but our result for @xmath66 has a ferromagnetic - like spin susceptibility which suppresses @xmath18 .
s. onari , r. arita , k. kuroki , and h. aoki , phys .
b * 68 * , 024525 ( 2003 ) .
m. takigawa , m. ichioka , k. kuroki , and y. tanaka , cond - mat/0504106 ( 2005 ) .
|
the pairing symmetry in the electron mechanism for superconductivity is explored when charge fluctuations coexist with spin fluctuations .
the extended hubbard model is adopted to obtain , with the fluctuation exchange approximation , a phase diagram against the on - site coulomb repulsion @xmath0 and the off - site repulsion @xmath1 for the square lattice with second - neighbor hopping @xmath2 .
we have found that ( i ) for large @xmath3 a triplet superconductivity with a @xmath4 symmetry can appear just below the charge density wave phase .
the pairing is degenerate with @xmath5 , so a chiral @xmath6 that breaks the time reversal symmetry should result , which is a candidate for the gap function on the @xmath7 band of sr@xmath8ruo@xmath9 and is consistent with a recent measurement of the specific heat .
( ii ) by systematically deforming the fermi surface with varied @xmath2 , we have identified the region where the triplet pairing is most favored to be the region where the fermi surface traverses the van hove singularity with the charge susceptibility strongly enhanced . 2
|
sn ia observations ( riess et al . 1998 ; perlmutter et al . 1999 ) confirmed that the observable universe is undergoing an accelerated expansion .
this acceleration is realized due to unknown cosmic fluid - dark energy ( de ) which have positive energy density and negative pressure .
so , it violate the strong energy condition ( sec ) .
the authors of ref . [ 3 ] confirmed that the violation of sec gives anti gravitational effect that provides an elegant description of transition of universe from deceleration zone to acceleration zone . in the literature ,
cosmological constant is the simplest candidate to describe the present acceleration of universe ( grn and hervik 2007 ) but it suffers two problems - the fine tunning and cosmic coincidence problems ( carroll et al .
1992 , copeland et al .
2006 ) . in the early of 21 century , some authors ( copeland et al .
2006 , alam et al .
2003 ) have developed few scalar field de models , namely the phantom , quintessence and k - essence rather than positive @xmath6 . in the physical cosmology , the dynamical form of de with an effective equation of state ( eos ) , @xmath7 ,
were proposed instead of positive @xmath6 ( komastu et al . 2009 , riess et al .
2004 , astier et al .
+ the substantial theoretical progress in string theory has brought forth a diverse new generation of cosmological models , some of which are subject to direct observational tests .
firstly the gravitational effect of cosmic strings are investigated by letelier ( 1979 , 1983 ) .
the present day observations of universe indicate the existence of large scale network of strings in early universe ( kible , 1976 , 1980 ) .
recently , pradhan et al . ( 2007 ) and yadav et al . (
2009 ) have studied string cosmological model in non - homogeneous space - time in which geometric strings were considered as source of matter .
after the big bang , the universe may have undergone so many phase transitions as its temperature cooled below some critical temperature as suggested by grand unified theories ( zeldovich et al .
1975 ; kibble 1976 , 1980 ; everett 1981 ; vilenkin 1981 ) . at the very early stage of evolution of universe ,
the symmetry of universe was broken spontaneously that rises some topologically - stable defects such as domain walls , strings and monopoles ( vilenkin 1981 , 1985 ) . among all the three cosmological structures , only strings give rise the density perturbations that leads the formation of galaxies .
pogosian et al .
( 2003 , 2006 ) have suggested that the cosmic strings are not responsible for cmb fluctuations and observed clustering of galaxies . in 2011 ,
pradhan and amirhashchi ( 2011 ) have considered time varying scale factor that generates a transitioning universe in bianchi -v space - time .
yadav et al .
( 2011 ) and bali ( 2008 ) have obtained bianchi - v string cosmological models in general relativity . the magnetized string cosmological models are discussed by some authors ( chakraborty 1980 ; tikekar and patel 1992 , 1994 ; patel and maharaj 1996 ; singh and singh 1999 ; saha and visineusu 2008 ) in different physical contexts .
recently goswami et al ( 2015 ) have studied @xmath6cdm cosmological model in absence of string . in this paper
we have established the existence of string cosmological model in anisotropic hecmann - suchuking space-time.the organization of the paper is as follows : the model and field equations are presented in section 2 . in section 3 , 4 @xmath5 5 we obtain the expressions for hubble constant , luminosity distance and apparent magnitude . section 6 deals the estimation of present values of energy parameters .
finally the conclusions of the paper are presented in section 7 .
we consider a general heckmann - schucking metric @xmath8 where a , b and c are functions of time only .
+ the energy - momentum tensor for a cloud of massive strings and perfect fluid distribution is taken as @xmath9 with @xmath10 and @xmath11 where @xmath12 is the four velocity vector , @xmath13 is the rest energy density of the system of strings , @xmath14 is the tension density of the strings and @xmath15 is a unit space - like vector representing the direction of strings .
+ in co - moving co - ordinates , @xmath16 choosing @xmath17 parallel to @xmath18 , we have @xmath19 if the particle density of the configuration is denoted by @xmath20 , then @xmath21 the einstein field equations are @xmath22 choosing co - moving coordinates , the field equations ( 8) in terms of line element ( 1 ) can be written as @xmath23 @xmath24 @xmath25 @xmath26 where @xmath27,@xmath28 and @xmath29 stand for time derivatives of a , b , c respectively .
the mass - energy conservation equation @xmath30 gives @xmath31 where @xmath27,@xmath28 and @xmath29 stand for time derivatives of a , b and c respectively .
subtracting ( 10 ) from ( 9),(11 ) from ( 10 ) and ( 11 ) from ( 9 ) we get @xmath32 @xmath33 @xmath34 subtracting(16 ) from ( 14 ) , we get @xmath35 this equation can be re - written in the following form @xmath36 this is simplified as @xmath37 this equation suggests the following relation ship among a b and c @xmath38 where @xmath39 , otherwise @xmath14 would be zero .
we can assume @xmath40 further integrating equation ( 15 ) , we get the first integral @xmath41 where k is an arbitrary constant of integration .
equation(13 ) simplifies as @xmath42 using equation(7 ) , this becomes @xmath43 if matter and string tension co - exist without much interaction , this equation splits into two parts .
@xmath44 @xmath5 @xmath45 this gives following expression for matter density @xmath20 and string tension @xmath14 for p=0(dust filled universe ) @xmath46 and @xmath47 the hubble s constant in this model is @xmath48 equations ( 9)-(12 ) and ( 20 ) are simplified as @xmath49 @xmath50 @xmath51 we see from equation(28 ) that @xmath14=0 for n=2 and other equations convert to the einstein s field equations of lrs bianchi type i model for perfect fluid without string + the average scale factor a of lrs bianchi type i model is defined as @xmath52 the spatial volume v is given by @xmath53 equations(27)-(29 ) can be re - written as @xmath54 @xmath55 we now assume that the string tension @xmath14 , cosmological constant @xmath6 and the term due to anisotropy also act like energies with densities and pressures as @xmath56 @xmath57 @xmath58 it can be easily verified that energy conservation laws ( 22 ) @xmath5 ( 23 ) holds separately for @xmath59 , @xmath60 @xmath5 @xmath61 @xmath62 @xmath63 the equations of state for matter , @xmath14 , @xmath64 and @xmath6 energies are as follows @xmath65 @xmath66 @xmath67 @xmath68 @xmath69 where @xmath70=0 for matter in form of dust , @xmath71 for matter in form of radiation
. there are certain more values of @xmath70 for matter in different forms during the coarse of evolution of the universe .
now we use the following relation between scale factor a and red shift z @xmath72 the suffix(0 ) is meant for the value at present time.the energy density @xmath13 comprises of following components @xmath73 integrations of energy equations(35 ) yield @xmath74 @xmath75 where suffix i corresponds to various energies densities .
we write equations ( 32 ) and ( 33 ) as @xmath76 @xmath77
the present stage of the universe is full of dust for which pressure @xmath78 .
we define critical density @xmath79 as @xmath80 then ( 41 ) gives @xmath81 where @xmath82 @xmath83 @xmath84 adding all the three above and using ( 43 ) , we get expression for hubble s constant @xmath85\\ & = & h^{2}_0\bigl[(\omega_m)_0(\frac{a_0}{a})^{n+1}(\omega_\lambda)_0(\frac{a_0}{a})^n+(\omega_\sigma)_0(\frac{a_0}{a})^{2(n+1)}+(\omega_\lambda)_0\bigr]\end{aligned}\ ] ] @xmath86 now we present two graphs on the basis of above equation(see fig . 1 & fig
these figures show that hubble s constant increases over red shift .
we can say that hubble s constant decreases over time .
we also notice from fig .
1 that in the case lambda dominated universe ( @xmath87=0 or .2820 ) hubble s constant vary slowly over time in comparison to matter dominated universe(@xmath88 ) .
if x be the spatially coordinate distance of a source with red shift z from us , the luminosity distance which determines flux of the source is given by @xmath89 geodesic for metric ( 1 ) ensures that if in the beginning @xmath90 then @xmath91 so if a particle moves along x- direction , it continues to move along x- direction always.if we assume that line of sight of a vantage galaxy from us is along x - direction then path of photons traveling through it satisfies @xmath92 from this we obtain @xmath93}}\end{aligned}\ ] ] @xmath94 where we have used @xmath95 and from eqns ( 26 ) and ( 37 ) @xmath96 .
+ so the luminosity distance is given by @xmath97}}\ ] ]
the absolute and apparent magnitude of a source ( m and m ) are related to the red shift of the source by following relation @xmath98 the type ia supernova ( sn ia ) can be observed when white dwarf stars exceed the mass of the chandrasekhar limit and explode .
the belief is that sn ia are formed in the same way irrespective of where they are in the universe , which means that they have a common absolute magnitude m independent of the red shift z. thus they can be treated as an ideal standard candle .
we can measure the apparent magnitude m and the red shift z observationally , which of course depends upon the objects we observe .
+ to get absolute magnitude m of a supernova , we consider supernova at very small red shift .
let us consider a supernova 1992p at low - red shift z = 0.026 with m = 16.08.for low red shift supernova , we have the following relation @xmath99 putting values z = 0.026 , m = 16.08 and @xmath100 from ( 50 ) , + equation ( 49 ) gives m for all sn ia as follows .
@xmath101 from equations ( 49 ) and ( 51 ) @xmath102 the equation(52 ) gives observed value of luminosity distance in term of observed value of apparent magnitude whereas the equation ( 48 ) gives the theoretically value of luminosity distance in terms of red shift on the basis of our model .
next , from equations ( 52 ) and ( 48 ) we get the following expression for ( m , z ) relations of supernova s sn ia .
@xmath103 @xmath104}}\ ] ]
string tension play important role in the study of early evolution of the universe .
we also notice that after publication of wmap data , today there is considerable evidence in support of anisotropic model of universe . in the past
, there might had been certain anisotropies in the universe .
our model validates that both string tension as well as anisotropy decreases over time .
considering it , we give very low present values to @xmath106 and @xmath107 . in our earlier work we developed models with n = 2 which make string tension @xmath14=0 .
so we take @xmath108 in our present model .
now we present two tables(see appendix-1 and 2 ) which contain @xmath109 high red shift ( @xmath110 ) sn ia supernova data of observed apparent magnitude and luminosity distances along with their possible error from union @xmath111 compilation .
these tables also contains the corresponding theoretical values of apparent magnitudes and luminosity distances for @xmath112 = @xmath113 , @xmath114 and @xmath115 respectively which are obtained as per from equation(48 ) and ( 52 ) .
as stated in the introduction , our purpose is to obtain the results close to wmap on the basis of union 2.1 compilations for our model , so we have obtained the various sets of theoretical data s of apparent magnitudes and luminosity distances corresponding to different values of @xmath112 in between @xmath114 to @xmath115 . in order to see that out of the these sets of theoretical data s of apparent magnitudes and luminosity distances which one is close to the observational values , we calculate @xmath116 using following formula due to amanullah et al.(2010 ) . the tables[1 ] and [ 2 ] describe the various values of @xmath116 against values of @xmath112 ranging in between @xmath114 to @xmath115 .
@xmath117 000 alam , u. , sahni , v , saini , t. d. , starobinsky , a. a. : mnras * 344 * , 1057 ( 2003 ) amanullah r. et al .
: astrophys .
j. * 716 * , 712 ( 2010 ) astier , p. , et al .
: astron . astrophys . *
447 * , 31 ( 2006 ) bali , r. : electron .
phys . , * 5 * , 105 ( 2008 ) chakraborty , s. : ind .
j. pure appl .
phys . , * 29 * , 31 ( 1980 ) carroll , s. m. , press , w. h. turner , e. l. : ann . rev . astron
. astrophys . * 30 * , 499 ( 1992 ) copeland , e. j. , sami , m. , tsujikawa , v. : int . j. mod
d * 15 * , 1753 ( 2006 ) @xmath120 .
gr@xmath121n , s. hervik : einstien s general theory of relativity : with modern application in cosmology , springer , new york ( 2007 ) goswami , g. k. , yadav , a. k. , mishra , m. : int .
* 54 * , 315 ( 2015 ) everett .
, a. e. : phys .
, * 24 * , 858 ( 1981 ) kibble , t. w. b. : j. phys . a : math .
* 9 * , 1387 ( 1976 ) kibble , t. w. b. : phys .
, * 67 * , 183 ( 1980 ) komastu , v , et al .
: astrophys .
. ser . * 180 * , 330 ( 2009 ) latelier , p. s. : phys .
d , * 20 * , 1294 ( 1979 ) latelier , p. s. : phys .
d , * 28 * , 2414 ( 1983 ) pradhan , a. , amirhaschi , h. : mod .
lett . a , * 26 * , 2261 ( 2011 ) pradhan , a. , yadav , a. k. , singh , r .p . and singh , v. k. : astrophys space sci . ,
* 312 * , 145 ( 2007 ) patel , l. k. , maharaj , s. d. : pramana j. phys .
, * 47 * , 1 ( 1996 ) perlmutter , s. , et al .
: astrophys .
j. * 517 * , 565 ( 1999 ) pogosian , l. , wasserman , i. and wyman , m. : arxiv : astro - ph/0604141 pogosian , l. , tye , s. h. , wasserman , i. and wyman .
m. : phys .
d , * 68 * , 023506 ( 2003 ) riess , a. g. , et al .
j. * 116 * , 1009 ( 1998 ) riess , a. g. , et al .
j. * 607 * , 665 ( 2004 ) singh , g. p. , singh , t. : gen .
, * 31 * , 371 ( 1999 ) saha , b. , visinescu , m. : int .
phys . , * 49 * , 1411 ( 2008 ) tikekar , r. , patel , l. k. : gen . rel .
grav . , * 24 * , 397 ( 1992 ) tikekar , r. , patel , l. k. : pramana j. phys .
, * 42 * , 483 ( 1994 ) vilenkin , a. : phys . rev .
d , * 24 * , 2082 ( 1981 ) vilenkin , a. : phys .
, * 121 * , 263 ( 1985 ) yadav , a. k. , yadav , v. k. and yadav , l. : int . j. theor .
phys . , * 48 * , 568 ( 2009 ) yadav , a. k. , yadav , v. k. and yadav , l. : pramana j. phys . ,
* 76 * , 681 ( 2011 ) zeldovich , ya .
b. , kobzarev , yu . , and okun , l. b. : zh .
fiz , * 67 * , 3 ( 1975 )
|
in the present work we have searched the existence of the late time acceleration of the universe with string fluid as source of matter in anisotropic heckmann - suchking space - time by using 287 high red shift @xmath0 sn ia data of observed absolute magnitude along with their possible error from union 2.1 compilation . it is found that the best fit values for @xmath1 , @xmath2 , @xmath3 and @xmath4 are 0.2820 , 0.7177 , 0.0002 @xmath5 -0.5793 respectively .
several physical aspects and geometrical properties of the model are discussed in detail .
+ + + + + + + + + * key words : * string , sn ia data , heckmann - suchking space - time .
|
quantum correlations arising in space - like separated measurement events can be subject to strong correlations that can not be explained by classical physics @xcite . the strength of classical correlations is bounded by bell inequalities , which are known to have quantum violations , certifying the non - classicality of the physics at play . a necessary condition to violate a bell inequality is sharing an entangled quantum state @xcite . for pure states ,
the standard measure of entanglement is the entanglement entropy @xcite , defined as the von neumann entropy of a subsystem of the state .
the relation between entanglement and the strength of quantum correlations has long been considered an interesting question .
for the simplest bell scenario , with two parties alice and bob performing one of two two - outcome measurements , the chsh inequality @xcite constitutes a tight bound on the set of classical correlations @xcite . to maximally violate the chsh inequality , one must distribute a state with maximal entanglement entropy ( the so - called maximally entangled state ) .
however , when the chsh inequality was generalized to the cglmp inequality @xcite which considers scenarios with alice and bob choosing one of two @xmath1-outcome measurements , it was discovered that the maximal quantum violation of the cglmp inequality for some low values of @xmath4 can not be achieved by distributing a maximally entangled state between alice and bob @xcite .
the cglmp inequality has been studied for scenarios with number of outcomes ranging up to the order of @xmath5 , and the most non - classical state is found to be far from maximally entangled @xcite .
recent results @xcite suggest that the discrepancy is due to the chosen measures quantifying quantum correlations .
quantum correlations are relevant in bell scenarios beyond the scope of the cglmp inequality .
significant research effort has been directed at bell scenarios where alice and bob can choose between more than two measurements .
such scenarios with two - outcome measurements for alice and bob have been explored in e.g. refs .
@xcite and many - outcome scenarios in e.g. refs .
in addition , bell scenarios with many settings and more than two parties have been studied in @xcite . here
we present a much simplified version of the multi - setting generalization of the cglmp inequality derived in ref .
@xcite i.e. for scenarios where alice and bob perform one of @xmath0 @xmath1-outcome measurements .
we investigate quantum violations of this inequality for arbitrary @xmath0 and @xmath1 , using both maximally entangled states and general entangled states .
in addition to the maximal violations of the inequality , we study the entanglement entropy of the most non - classical quantum state by both analytical and numerical means .
interestingly , we find that the nature of the most non - classical state is very different depending on whether @xmath2 ( the cglmp inequality ) or @xmath6 . when @xmath2 , the entanglement entropy of the most non - classical state is a monotonically decreasing function of @xmath1 and thus as we increase @xmath1 the state becomes less entangled .
however , we find that when @xmath6 the entanglement entropy is a non - monotone function of @xmath1 and that the entanglement entropy of the most non - classical state converges to that of the maximally entangled state in the limit of large @xmath1 .
let alice and bob share a bipartite quantum state of local hilbert space dimension @xmath7 : @xmath8 .
alice and bob can each perform one of @xmath0 measurements , indexed by @xmath9 respectively , with associated outcomes @xmath10 each of which can take @xmath1 different values , where @xmath11 , on some ordered space .
we will focus on cases where the number of outcomes equals the local hilbert space dimension ; that is when @xmath12 .
the constraint @xmath12 will from now on be assumed unless otherwise is stated .
scenarios with @xmath13 will be briefly discussed . for the given bell scenario
, we can construct the following bell inequality , @xmath14 the inequality is a chained bell inequality , and we illustrate it graphically in figure [ fig : ineq ] . to prove this assertion , we extend the method of ref .
first note the following fact : @xmath15 .
therefore , if we take the complement of both sides , it holds that @xmath16 ( see figure [ fig : ineq ] ) .
any classical probability distribution in the given bell scenario takes the form of a convex mixture of fully deterministic probability distributions , for which all probabilities are trivial i.e. either zero or one .
therefore , since randomness is not a fundamental property but a mere statistical property in classical probability distributions , it must hold that @xmath17 , from which the inequality follows . ' between two measurement outcomes . ]
our inequality is ( up to a permutation of parties and measurement settings ) a simplified version of the inequality derived in @xcite , analogous to the simplification of the cglmp inequality in ref .
an advantage of our inequality is that its form is independent of @xmath1 .
another advantage is that the inequality only depends on the relative ordering of the outcomes , and not their ascribed values . for sake of completeness
, we show that of the specific case when the outcomes take the values @xmath18 , our inequality is equivalent to that of ref .
@xcite ( again , up to a permutation of parties and measurement settings ) , namely @xmath19\rangle+\langle [ a_0-b_1]\rangle+ ... +\langle [ b_{n-1}-a_{n-1}]\rangle\\ + \langle [ a_{n-1}-b_0 - 1]\rangle\geq d-1\end{gathered}\ ] ] where @xmath20 and @xmath21=x \mod{d}$ ] .
simply note that we can write @xmath21=x - d\lfloor x / d\rfloor$ ] which implies that the inequality can be written @xmath22 .
if we use that @xmath23 , the inequality is implied .
we now investigate the quantum properties of the bell expression @xmath24 .
we consider both violations using maximally entangled states i.e. a state of the form @xmath25 , and non - maximally entangled states .
let us denote the outcomes of alice s and bob s measurements by @xmath26 . in ref .
@xcite , particular measurements were introduced with the motivation that the maximally entangled state has the property that if alice measures an observable with an eigenstate @xmath27 , and bob measures an observable with conjugated eigenvectors @xmath28 , correlations will be perfect .
therefore , we let alice and bob measure in the respective bases given by @xmath29 and @xmath30 for @xmath9 , and @xmath31 .
we have conducted extensive numerical studies using semidefinite programs ( sdps ) @xcite to investigate whether these measurements are optimal .
however , the problem of optimizing the quantity @xmath24 over the measurement of alice , the measurements of bob , and the shared state , can not be directly maped into an sdp . instead we have used a see - saw technique in which we first use an sdp to optimize only over the measurements of alice .
then we run another sdp to optimize over the measurements of bob while fixing alice s measurements as obtained in the previous sdp .
finally , we fix alice s and bob s measurements as returned by the two previous sdps and run a third sdp optimizing over only over the shared state .
this procedure is iterated until a stable value of @xmath24 is obtained .
naturally , this procedure does not guarantee that @xmath24 achieves its global optimum under quantum theory .
yet , for all investigated cases ranging over several low values of @xmath32 , the obtained value of @xmath24 by the see - saw method returns the above measurements of alice and bob . due to this numerical evidence
, we will conjecture the above measurements as optimal .
the numerics supporting the conjecture of the optimal measurements is performed with the constraint that @xmath12 .
we have used sdps in an analogous see - saw technique to investigate the value of @xmath24 in various scenarios when @xmath13 .
specifically , we have considered @xmath33 , @xmath34 and @xmath35 .
we have observed no instance in which @xmath13 yields an enhancement over the case @xmath12 . following the conjecture of the optimal measurements
, we can define the bell operator , @xmath36 , associated with the bell expression @xmath24 as @xmath37 thus , we can write that @xmath38 where @xmath39 is some shared quantum state . the optimal value of @xmath24 is found from computing the smallest eigenvalue of @xmath36 , and the corresponding eigenvector is the associated quantum state that gives rise to the maximal violation i.e. it is the most non - classical quantum state .
however , in the case of pure quantum states we can extensively simplify the eigenvalue computation by using the schmidt decomposition to write any pure quantum state in the form @xmath40 for @xmath41 being real numbers respecting the normalization @xmath42 .
it is then possible to write the bell expression @xmath24 in the following form @xmath43 where the @xmath44 matrix @xmath45 is given by @xmath46 this expression is derived in appendix [ appa ] .
the problem of finding the maximal quantum violation of the inequalities reduces to finding the smallest eigenvalue of @xmath45 .
however , we will first investigate optimal values of @xmath24 achievable by performing measurements on a shared maximally entangled state .
the maximally entangled state , @xmath47 has schmidt coefficients @xmath48 for @xmath49 . inserting this into
, the value of the bell expression is easily computed from @xmath50 let us study the limiting case of many measurements , i.e. find the value of @xmath51 .
we first compute the probability distribution arising from alice and bob making measurements and on @xmath47 .
this can be obtained directly from equation : @xmath52 using the probability distribution , we can expand the bell expression @xmath53 in as @xmath54.\end{gathered}\ ] ] in the limit of large @xmath0 , it is evident that there is a contribution to @xmath55 only from the final term in the sum associated to @xmath56 . using taylor expansion of the sine function
, we find that the limit is @xmath57 which goes to zero as @xmath58 .
in fact , the right - hand - side of is the no - signaling bound of @xmath24 .
this is realized from the fact that the no - signaling bound of @xmath59 in coincides with its algebraic bound , that is zero , and that the affine transformation relating @xmath24 and @xmath59 is @xmath60 up to permutations of parties and settings .
thus , in the limit of many measurement settings , the above result of quantum theory does converge to the no - signaling bound .
however , the algebraic bound of @xmath24 is zero , and thus there is a discrepancy between the no - signaling bound and the algebraic bound of @xmath24 . in the other limiting case of many outcomes , finding
the value of @xmath61 is a bit more involved .
we can use the fact that the sum in equation can be evaluated as an integral , which can be related to the trigamma function , @xmath62 , defined as the second logarithmic derivative of the gamma function , @xmath63 , as follows @xmath64 in appendix [ appb ] , we prove this relation .
it is worth noting that for the special case of @xmath2 , it was shown in @xcite that the limit evaluates to @xmath65 where @xmath66 is catalan s constant , corresponding to what was obtained for the cglmp inequality in ref .
@xcite .
evidently , unless we let both @xmath67 , we can not reach the algebraic bound of @xmath24 by using maximally entangled states .
this puts our inequality in contrast to that of ref.@xcite in which the algebraic bound coincides with the no - signaling bound and thus can be asymptotically achieved in the limit @xmath68 .
we will now investigate the optimal quantum violations of the inequalities by sharing general entangled states .
as we have already mentioned , the optimal quantum violation of the inequality corresponds to the smallest eigenvalue of the matrix @xmath45 and the entangled state giving rise to the violation is the associated eigenvector of @xmath45 .
this reasoning relies on the measurements and indeed being optimal choices . to strengthen this conjecture ,
we have verified the optimality of these results numerically using sdps @xcite for the cases @xmath69 .
and dimensions up to about @xmath70 .
solid lines are obtained from maximally entangled states , the dotted lines from the most non - classical quantum state , and the dashed lines from the approximation state @xmath71 in . ] in figure [ fig:1 ] , we have computed the maximal violation of the inequalities , the violation using a maximally entangled state , and also the violation due to the approximate state @xmath71 which is discussed below .
we have chosen some particular values of @xmath72 and @xmath1 ranging from @xmath73 up to approximately about @xmath70 in exponentially increasing increments .
the computation of the smallest eigenvalue of @xmath45 ( the maximal violation ) can be efficiently implemented using power iteration over the krylov space associated to @xmath45 .
we elaborate on the numerical technique in appendix [ appc ] .
the results suggest that the optimal violation of the inequality approaches the algebraic bound of @xmath24 , namely zero , as @xmath74 for all investigated values of @xmath0 .
the convergence to the algebraic bound is more rapid with increasing @xmath0 . due to the apparent discrepancy between
maximally entangled state and the most non - classical state , @xmath75 , in terms of their ability to minimize @xmath24 , it is interesting to investigate the properties of @xmath75 . for simplicity ,
if we represent @xmath76 in the schmidt basis as @xmath77 , it follows from the symmetries of the matrix @xmath45 that @xmath75 is subject to the symmetry property that @xmath78 whenever @xmath79 .
however , despite being easy to compute for given values @xmath80 , the exact form of @xmath76 is difficult to find by analytical means .
however , we propose an approximation of the state @xmath75 for arbitrary @xmath0 and @xmath1 , which denote by @xmath71 : @xmath81^{-\frac{2}{n}}$ ] is the normalization .
the special case amounting to @xmath2 was initially proposed in ref .
@xcite to approximate the state maximally violating the cglmp inequality .
the state @xmath82 was extensively studied for dimensions up to the order of @xmath5 in ref .
@xcite . on the one hand
, it was shown that @xmath82 yields a good approximation of the optimal violation of the cglmp inequality , especially for @xmath83 , and that it yields a violation which converges to the algebraic bound in the limit of many outcomes @xcite . on the other hand , the numerics of ref .
@xcite shows that @xmath82 is quite a bad approximation of the state @xmath75 in terms of entanglement entropy .
although it was shown that the entropy of @xmath82 becomes @xmath84 in the limit of many outcomes , where it converges to the most non - classical state @xcite .
for general @xmath0 , the numerics in figure [ fig:1 ] indicate that the violation due to the approximate state @xmath71 converges to the algebraic bound of @xmath24 in the limit of many outcomes .
furthermore , it appears that with higher @xmath0 , the violation due to @xmath71 increasingly well approximates the optimal quantum violation of the inequality .
the question of how well the state @xmath71 approximates the entanglement entropy of the most non - classical state @xmath75 is investigated in the next section .
in order to investigate the relation between maximal non - classicality and entanglement , we will study the entanglement entropy of the state @xmath75 by extensive numerics for high @xmath0 and @xmath1 .
we will also study to what extent the degree of entanglement of @xmath75 can be approximated with the introduced state @xmath85 .
( dashed line ) for @xmath72 and dimensions up to @xmath70 . ] using the schmidt basis representation , the entanglement entropy of a state @xmath86 is given by @xmath87 which is easily realized to be a number between zero and one ( see note .
division by @xmath88 means measuring entropy in units of bits . for @xmath1-level system ,
division by @xmath89 corresponds to measuring entropy in units of `` dits '' .
these are the natural units which normalizes the maximal value of entropy to be 1 . ] for conventions of units ) .
the entanglement entropy is maximized for the maximally entangled state for which @xmath90 . in figure
[ fig:2 ] we plot the entanglement entropy of the state @xmath75 and the approximate state @xmath71 for a few values of @xmath0 and for dimensions up to about @xmath70 .
remarkably , our results show that for the investigated values of @xmath0 exceeding @xmath2 , the entanglement entropy of the most non - classical state is not monotonic with @xmath1 .
the sharpness of the trough of the entanglement entropy increases with the value of @xmath0 .
our numerics suggests that @xmath91 is a non - monotonic function whenever @xmath92 .
in addition , while for small values of @xmath1 the state @xmath82 provides a less good approximation of the entanglement entropy of the most non - classical state @xcite , we see that for @xmath93 , @xmath94 becomes a significantly better approximation of the entanglement entropy of @xmath75 .
in fact , the accuracy of the approximate state with respect to @xmath75 appears to increase with both @xmath0 and @xmath1 . due to the very slow convergence of the entanglement entropy of @xmath75 , especially for fix @xmath0 and variable @xmath1 , it is hard to tell how it behaves for @xmath1 beyond the range we have studied .
however , it is reasonable to assume that as @xmath74 , the entanglement entropy converges to that of the state @xmath71 . in appendix
[ appd ] , we show the following limit @xmath95 evidently , the entanglement entropy behaves very differently depending on whether we consider case of the cglmp inequality ( @xmath2 ) or allow for more measurements .
also , this result implies that the entanglement entropy ( at the very least for the state @xmath71 ) is a non - monotonic function of @xmath1 whenever @xmath6 , which falls in line with our numerical results in figure [ fig:2 ] .
the above results show that @xmath96 entanglement entropy of the optimal and approximate state is non - monotonic for @xmath6 and that @xmath97 it converged to that of the maximally entangled state for @xmath6 .
note however , that this does not imply that the optimal state or approximate state are the maximal entangled state for @xmath6 and @xmath58 .
it simply means that the difference between their corresponding entanglement entropies becomes infinitesimal .
an intuitive picture can be obtained by plotting the corresponding schmidt coefficients @xmath98 as a function of @xmath99 , as was done in @xcite .
one observes that the shape of the curve is @xmath100-shaped and that for increasing @xmath1 , as well as @xmath0 the middle part becomes flatter and flatter .
the above result suggests that as @xmath101 the function @xmath102 remains @xmath100-shaped over @xmath103 $ ] for @xmath2 , while for @xmath6 it becomes essentially a constant function for @xmath104 , as in the case of the maximally entangled state , apart from a peak with infinitesimal support at @xmath105 and @xmath106 .
in fact , those two peaks with infinitesimal support make the maximal violations of the bell inequality for this state differ from that of the maximally entangled state .
note that entanglement entropy as defined in is the commonly used measure for entanglement of a state .
however , we can also look at the relative entropy or kullback - leibler divergence between the maximally entangled state and the optimal or approximate state , i.e. @xmath107 where @xmath86 is either the optimal or the approximate state . , between the maximally entangled state @xmath47 and @xmath86 , which for solid lines represents the optimal state @xmath75 and for dashed lines represents the approximate state @xmath108 . ]
this is shown in figure [ fig.relent ] as a function of @xmath1 for different values of @xmath0 .
furthermore , it is shown in appendix [ appd ] that @xmath109 with @xmath110 .
note that the divergence for @xmath2 is very slow , @xmath111 .
this expression agrees with the numerical results shown in figure [ fig.relent ] .
furthermore , it is in accordance with the above picture , where we can attribute the divergence of the relative entropy for @xmath2 to a finite difference of the @xmath98 and @xmath112 for infinitely many @xmath99 , corresponding to a finite interval of @xmath103 $ ] , while the finite value of the relative entropy for @xmath6 reflects the fact that @xmath98 and @xmath112 differ only for finitely many @xmath99 , corresponding to the two peaks at @xmath105 and @xmath106 which have infinitesimal support . to complete the discussion , we plot @xmath113 as a function of @xmath0 in figure [ fig.infkl ] .
one sees that the relative entropy between the maximally entangled state and the approximate states for infinite @xmath1 goes to zero as @xmath114 , which is in accordance with the results of @xcite , meaning that the peaks of @xmath115 at @xmath105 and @xmath106 become smaller as @xmath0 increases until for @xmath114 the approximate state becomes identical to the maximally entangled state .
in this work , we have introduced a much simplified version of the mutli - setting generalization of the cglmp inequality first presented in ref .
we have conjectured the optimal measurements and provided numerics supporting this claim for some low values of @xmath80 .
then , we studied quantum violations of our inequalities using both maximally entangled and generally entangled states and demonstrated that the maximal violations are obtained by non - maximally entangled states .
we gave strong numerical evidence that the tsirelson bound of the bell expression @xmath24 coincides with the algebraic bound of that quantity , namely zero .
however , our main result is the strong evidence for the non - monotonicity of the entanglement entropy of the most non - classical state , as a function of @xmath1 for @xmath6 . and the approximate state @xmath85 as a function of the number of the number of settings @xmath0 in the limit @xmath74 . ]
we see that the most non - classical state for any @xmath6 become more and more entangled with increasing @xmath1 and its entanglement entropy converges to that of the maximally entangled state .
this is very much opposite to the quantum behavior observed for the cglmp inequality @xcite i.e. when @xmath2 .
our results suggest that the discrepancy between entanglement and the strength of quantum correlations is strongly dependent on the number of settings of the bell scenario and is eradicated in the limit of large @xmath1 , given more than two settings .
finally , recent work @xcite studying the ability of random pure states to violate the cglmp inequality has shown that with increasing @xmath1 , the inequality is increasingly violated on average .
extending such an analysis to the case of @xmath6 settings would be of interest .
we thank richard gill for discussions .
also , we would like to acknowledge the use of the advanced research computing ( arc ) , university of oxford in carrying out this work .
s.z . acknowledges support by nokia technologies , lockheed martin and the university of oxford through the quantum optimisation and machine learning ( quopal ) project .
m.p . was funded by fnp prgramme team and ncn grant 2014/14/e / st2/00020 .
here we derive bell operator , @xmath45 , from equation , associated to the measurements and . given any shared quantum state written in the schmidt basis as @xmath40 , the associated probability distribution @xmath116 can be simplified to @xmath117 we can use the probability distribution to evaluate the bell expression in equation . if we additionally observe that @xmath118 , @xmath119 and @xmath120 , one sees that all probabilities appearing in are independent of the measurement choices . by direct insertion
, we write the bell expression as @xmath121\end{gathered}\ ] ] where the bell operator @xmath45 appears in the square brackets ( scaled by @xmath122 ) . in case of @xmath123 ,
corresponding to the diagonal elements of @xmath45 , it is easily found that @xmath124 to evaluate the off - diagonal elements , corresponding to @xmath125 , we first use the geometric series to evaluate each of the three series appearing inside the bracket of .
combining the exponentials in trigonometric functions leads to the final expression @xmath126\ ] ] where one identifies the matrix @xmath45 with the term inside the bracket .
here we compute the limit @xmath127 by evaluating the integral in equation . to compute the integral , we use the dirichlet series for the cosecant function , namely @xmath128 if we define @xmath129 , we obtain an expansion for one of the two factors appearing in the integrand of equation in terms of the variable @xmath130 . additionally inferring the notation @xmath131 and @xmath132 , we can write the integral expression in as a sum over an infinite series of integrals over integrands of the form @xmath133 . using euler s formula , we can write each integral in the infinite series as a sum of four elementary integrals which can easily be solved ; each over a term of the form @xmath134 or @xmath135
. however , two of the four integrals will vanish : it is a straightforward computation to find that @xmath136 for all values of @xmath137 .
the task of computing @xmath138 is reduced to evaluating the following series of integrals : @xmath139 $ ] .
the two elementary integrals appearing in this expression are evaluated to @xmath140 where we have defined @xmath141 and @xmath142 . using this expression
, one has @xmath143 the first summation , which we denote @xmath144 , can be re - written as @xmath145 .
recall that the trigamma function , @xmath146 , can be defined through the series @xmath147 , which is precisely the form of the summation in @xmath144 .
if we additionally use the recurrence relation @xmath148 , @xmath144 can be simplified as @xmath149 a similar analysis can be done for the second sum in , now denoted @xmath150 . first using the recurrence relation , and then using the reflection property of the trigamma function , @xmath151 , allows one to write @xmath152 plugging the final form of @xmath144 and @xmath150 back into , yields @xmath153 which is the final expression .
in this paper we have numerically computed both the maximal violation of the inequality and the entanglement entropy of the associated most non - classical state for values of @xmath1 ranging up to about @xmath70 , for a few values of @xmath0 .
this was achieved by using similar techniques as in @xcite which we will elaborate on here .
as previously mentioned , the maximal violation of the inequality is found from computing the smallest eigenvalue of the matrix @xmath45 , and the most non - classical state is the associated eigenvector @xmath154 of @xmath45 . recall that the components of this eigenvector @xmath154 are just the schmidt coefficients .
since @xmath45 is a @xmath44 matrix , the computational complexity of the problem scales as @xmath155 with increasing @xmath1 , becoming less and less feasible .
nevertheless , there exist more efficient numerical techniques for solving the problem , especially since we are only interested in the smallest eigenvalue of @xmath45 and the associated eigenvector .
we have used the method of power iteration over a krylov subspace which is the essence of the arnoldi and lanczos algorithm @xcite .
the krylov subspace of order @xmath156 associated to some @xmath157 matrix @xmath158 and some @xmath159 vector @xmath160 is the space spanned by the image of @xmath160 under the first @xmath161 powers of @xmath158 . for large @xmath161 , these images of @xmath160 approach the eigenvector of @xmath158 associated to its _ largest _ eigenvalue ( in modulus ) . for our purposes ,
we have fixed @xmath162 and started the iteration from the schmidt coefficients associated with the maximally entangled state i.e. @xmath163 . however , since we are interested in the _ smallest _ eigenvalue of @xmath45 , we have decomposed @xmath45 as @xmath164 where @xmath165 is the @xmath1-dimensional identity matrix .
we now exploit the simple fact that the eigenvectors of @xmath45 are the same as the eigenvectors of @xmath166 and that the corresponding eigenvalues are related by @xmath167 , in particular for @xmath168 being the smallest eigenvalue of @xmath45 and @xmath169 being the largest eigenvalue of @xmath166 .
the above described algorithm is summarized as follows : initialize : @xmath170 @xmath171 normalize : @xmath172 @xmath173 @xmath174 .
importantly , to avoid memory problems , we only store the current vector @xmath175 in memory when running the iteration . in addition , since @xmath45 has a banded structure we can store its independent elements in a @xmath176-dimensional vector .
further optimization can be achieved by exploiting the symmetry of @xmath45 or @xmath166 .
the program is implemented in c++ and ran on the arcus cluster at the advanced research computing , university of oxford .
data and code are available on request .
in this appendix we derive the limit .
that is , we evaluate @xmath177 where the schmidt coefficients are given by @xmath178^{\frac{1}{n}}}\ ] ] and @xmath179^{-\frac{2}{n}}$ ] is the normalization .
we first compute the normalisation to leading order by approximating the sum as an integral . for convenience ,
we introduce @xmath180 and the rescaled variable @xmath181 .
then the normalization can be written as @xmath182^{\frac{2}{n } } } + ... , \ ] ] where the dots refer to next - to - leading - order terms .
explicit evaluation of the integral yields to leading order @xmath183 where @xmath184 which is finite for @xmath93 .
note that the explicit form of @xmath185 is not relevant for this derivation , but will be relevant later .
approximating the sum in the entanglement entropy in a similar manner one arrives at @xmath186}{\left[(x+\epsilon)(1-x)\right]^{\frac{2}{n}}}dx + .... \end{gathered}\ ] ] the integral in the second term can be evaluated as follows for @xmath2 @xmath189}{\left[(x+\epsilon)(1-x)\right]^{\frac{2}{n}}}dx = - |\log\epsilon|^2 + ... \label{eq : int1}\ ] ] while for @xmath93 one has
@xmath189}{\left[(x+\epsilon)(1-x)\right]^{\frac{2}{n}}}dx = \mathcal{o}(\epsilon^{1 - 2/n } |\log \epsilon| ) \label{eq : int2}\ ] ] combining this with , we see that the second term in becomes @xmath190}{\left[(x+\epsilon)(1-x)\right]^{\frac{2}{n}}}dx \\
= \begin{cases } -\frac{1}{2}+ ... & \text{$n=2 $ } , \\
\mathcal{o}(\epsilon^{1 - 2/n } ) & \text{$n\geq 3 $ } , \end{cases } \label{eq : second - term}\end{gathered}\ ] ] which for @xmath93 becomes zero in the limit @xmath187 . plugging and into and taking the limit finally gives @xmath191 which completes the derivation . for completeness
we also calculate the relative entropy or kullback - leibler divergence between the maximally entangled state and the approximate state in the limit where the number of outcomes goes to infinity .
that is @xmath192 following a similar strategy as above one obtains @xmath193 dx + ...
\nonumber\\ & = & \log\left ( \mathcal{n}\epsilon^{1-\frac{4}{n } } \right ) + \frac{2}{n } ( -2 + 2\epsilon -2\epsilon\log\epsilon ) + ... \end{aligned}\ ] ] in the limit where @xmath58 , or equivalently @xmath194 , the second term converges to @xmath195 .
furthermore , we see from that the first term diverges for @xmath2 , while it converges to @xmath196 for @xmath92 . in total
one has @xmath109 where @xmath185 is given in .
note that the divergence for @xmath2 is very slow . indeed from the above expression
, one sees that it diverges as @xmath111 when @xmath101 .
s. braunstein and c. caves , writing out better bell inequalities , ann .
( ny ) , * 202 * , 22 ( 1990 ) . ; n. gisin , bell inequality for arbitrary many settings of the analyzers , phys . lett .
a , * 260 * , 1 ( 1999 ) . ;
i. pitowsky and k. svozil , optimal tests of quantum nonlocality , phys .
a * 64 * , 014102 ( 2001 ) . ;
n. brunner and n. gisin , partial list of bipartite bell inequalities with four binary settings , phys .
lett . a , * 372 * , 3162 ( 2008 ) . ;
k. f. pal and t. vertesi , quantum bounds on bell inequalities , phys .
a * 79 * , 022120 ( 2009 ) ; k. f. pal and t. vertesi , maximal violation of a bipartite three - setting , two - outcome bell inequality using infinite - dimensional quantum systems , phys . rev .
a * 82 * , 022116 ( 2010 ) .
ji , j. lee , j. lim , k. nagata , and h - w . lee , multisetting bell inequality for qudits , phys . rev .
a * 78 * , 052103 ( 2008 ) . ; m. zukowski and a. datta , geometric chained inequalities for higher - dimensional systems , phys . rev .
a * 90 * , 012106 ( 2014 ) .
|
the discrepancy between maximally entangled states and maximally non - classical quantum correlations is well - known but still not well understood .
we aim to investigate the relation between quantum correlations and entanglement in a family bell inequalities with @xmath0-settings and @xmath1 outcomes . using analytical as well as numerical techniques , we derive both maximal quantum violations and violations obtained from maximally entangled states .
furthermore , we study the most non - classical quantum states in terms of their entanglement entropy for large values of @xmath1 and many measurement settings .
interestingly , we find that the entanglement entropy behaves very differently depending on whether @xmath2 or @xmath3 : when @xmath2 the entanglement entropy is a monotone function of @xmath1 and the most non - classical state is far from maximally entangled , whereas when @xmath3 the entanglement entropy is a non - monotone function of @xmath1 and converges to that of the maximally entangled state in the limit of large @xmath1 .
|
for many years , it was thought that fermi statistics were fundamental , in the sense that one could only obtain a theory with fermionic excitations by introducing them by hand ( via anti - commuting fields ) .
then , over the past two decades , this view began to change .
a number of real world and theoretical examples showed that fermions and anyons could _ emerge _ as low energy collective modes of purely bosonic systems .
the first examples along these lines were the fractional quantum hall states .
@xcite usually we think of the fqh states as examples of anyonic excitations emerging from interacting fermions .
however , from a purely theoretical point of view , the same effects should occur in systems of interacting bosons in a magnetic field.@xcite there were also indications of emerging fermions in the slave - boson approach to spin-1/2 systems@xcite and in the study of resonating valence bond ( rvb ) states.@xcite unfortunately , the rvb picture and the slave boson approach both rely on approximate or mean - field techniques to construct and analyze these exotic states .
more recently , a number of researchers have introduced exactly soluble or quasi - exactly soluble models with emerging fermions .
@xcite these models allow for a more well - controlled analysis , albeit in very specific cases . the mean - field approach and the exactly - soluble examples both provide clues to the structure and basic properties of bosonic models with fermionic excitations .
they indicate , among other things , that fermions never appear alone in lattice spin systems .
instead they always come together with a nontrivial gauge field and the emerging fermions are associated with the deconfined phase of the gauge field.@xcite the deconfined phases always seem to contain a new kind of order - topological order,@xcite and the emerging fermions and anyons are intimately related to the new order .
this was particularly apparent in the context of the slave - boson approach . in this technique
, one expresses a spin-1/2 hamiltonian in terms of fermion fields and gauge fields.@xcite clearly , the presence of fermion fields does not , by itself , imply the existence of fermionic quasi - particles - the fermions appear only when the gauge field is in the _ deconfined _ phase .
constructed several deconfined phases where the fermion fields do describe well defined quasi - particles .
depending on the properties of the deconfined phases , these quasi - particles can carry fractional statistics ( for the chiral spin states ) @xcite or fermi statistics ( for the @xmath2 deconfined states).@xcite although it was less evident initially , a similar picture has emerged from the study of rvb states and the associated quantum dimer models . originally proposed that fermions ( spinons ) could emerge from a nearest neighbor dimer model on a square lattice .
later , it was realized that the fermions couple to a @xmath2 gauge field , and the fermionic excitations only appear when the field is in the deconfined phase.@xcite it turns out that the dimer liquid on square lattice with only nearest neighbor dimers is not in the deconfined phase ( except at a critical point).@xcite however , on a triangular lattice , the dimer liquid does have a @xmath2 deconfined phase.@xcite the results in are valid in this case and fermionic quasi - particles do emerge in a dimer liquid on a triangular lattice . in this paper
, we attempt to clarify these observations and to put them on firmer foundations .
we give a general argument which shows that emerging fermions always occur together with a nontrivial deconfined gauge field .
in addition , we derive an algebraic formula that allows one to calculate the statistics of a lattice particle from the properties of its hopping operators .
we feel that this formula both elucidates the fundamental meaning of statistics in a lattice system and simplifies their computation .
this approach has the added advantage that it works in any number of dimensions - unlike the flux binding picture@xcite developed in fqh theory .
we would like to point out that all the previously mentioned examples of emerging fermions are two dimensional models .
the emerging fermions in those models are related to the flux binding picture in .
our algebraic approach allows us to establish the emergence of fermions in an exactly soluble 3d bosonic model .
our paper is organized as follows : in the first section , we give a definition of the statistics of a lattice particle . in the second section
, we derive the algebraic statistics formula discussed above . in the third section , we apply the formula to the case of fermionic ( or anyonic ) excitations in a lattice spin system .
the formula demands that the fermions always come in pairs , and that their pair creation operator has a string - like structure with the newly created particles appearing at the ends .
we show that the strings represent gauge fluctuations , and the fermions carry the corresponding gauge charge . after presenting the general argument
, we devote the last two sections to exactly soluble examples of this phenomenon in @xmath0 and @xmath1 dimensions .
the above result suggests a very interesting picture of emerging fermions .
fermions and gauge fields appear to be two sides of the same coin : in some sense , fermions are the ends of strings and gauge fields are the fluctuations of strings
. extended string - like structures seem to be the key to understanding both fermi statistics and gauge fields .
what do we mean when we say that a particle is a `` boson '' or a `` fermion '' ?
the most common way to define statistics is to use the algebra of creation and annihilation operators .
if these operators satisfy the bosonic algebra , the corresponding particles are said to be bosons ; if they satisfy the fermionic algebra , the particles are said to be fermions . in this paper , we will use another , equivalent , definition based on the path integral formulation of quantum mechanics . given any @xmath3 particle system with short - range interactions
, we can write down the corresponding multi - particle action @xmath4 . the most general action @xmath4 has two components : @xmath5 the first term , @xmath6 , can be any local expression .
typically , it s just the usual classical action .
the second term , @xmath7 , is less familiar . in general
, it can be any expression which only depends on the topology ( e.g. homotopy class ) of paths . in @xmath1 or more dimensions , the form of @xmath7
is highly constrained . the amplitude for a closed path @xmath8 must be of the form @xmath9 where @xmath10 is the number of particle exchanges that occur in @xmath8 .
the statistics of the particles are defined by the sign in this expression .
if this sign is positive , we say the particles are``bosons '' , if the sign is negative , the particles are `` fermions '' . in @xmath0 dimensions , there are other , more complicated possibilities for @xmath7 .
these particles are called `` anyons . ''
according to this definition , the statistics are completely determined by the topological term @xmath7 in the action .
one way to isolate this term is to compare the amplitude for two paths which are the same locally ( this will be made more precise in the next section ) but differ in their global properties . in particular , we can compare the amplitude for a path which exchanges two particles with another path which is the same locally , but does nt exchange the two particles .
the difference between the phases of the two amplitudes is then precisely @xmath11 , the statistical phase .
we now reformulate this ( theoretical ) test of statistics so that it can be applied to particles on a lattice .
the prescription is as follows : we take a two particle state @xmath12 , and consider a product of hopping amplitudes along a lattice path which exchanges the particles @xmath13 each hopping amplitude in this expression involves one of the two particles moving to a neighboring site while the other particle remains fixed . at the end
, the two particles have exchanged places : @xmath14 .
now we construct another path which is the same locally , but does nt exchange the two particles , and take a product of hopping amplitudes along this path .
the claim is that the difference between the phases of these two expressions is precisely the statistical phase of the particles .
that is , the particles are bosons , fermions , or anyons depending on whether the phase difference is @xmath15 , @xmath16 , or something else . one way to see this is to remember the derivation of the path integral formulation of quantum mechanics .
according to the standard derivation , the amplitude @xmath17 for a two particle path @xmath18 is given by a product @xmath19 in the limit that @xmath20 .
now , in the discrete ( lattice ) case , the above expression can be further simplified . since @xmath21 , we can rewrite it as @xmath22 successive states @xmath23 must either be identical , or
must differ by a single particle hop .
the matrix elements between identical states do nt contribute to the phase .
thus , we can drop them without affecting our result .
we are left with : @xmath24 where the @xmath25s are all distinct .
now , as we discussed earlier , the statistical phase can be obtained by comparing the phase of this amplitude with another amplitude which is the same locally but does nt exchange the two particles .
when we make this comparison , the factors of @xmath26 drop out ( since they contribute equally to the two products ) .
thus , it suffices to compare products of the form : @xmath27 this is precisely the expression in .
in this section we derive a simple algebraic formula for the statistics of a particle hopping on a lattice .
this formula is completely general and holds irrespective of whether the particles are fundamental or are low energy excitations of an underlying condensed matter system ( e.g. quasi - particles ) .
we begin with a hilbert space which describes @xmath10 hard - core particles hopping on a @xmath28 dimensional lattice .
the states can be labeled by listing the positions of the @xmath10 particles : @xmath29 .
the particles are identical so the states @xmath29 do not depend on the order of @xmath30 .
for example , @xmath31 a typical hamiltonian for this system is of the form @xmath32 where @xmath33 are `` hopping operators '' with the property that @xmath34 we assume that the hopping is local , _ i.e _ @xmath35=0 \label{loc}\ ] ] if @xmath36 are all different .
our goal is to compute the statistics of the particles described by this hopping hamiltonian @xmath37 . in the following ,
we show that the statistical angle can be derived from the simple algebraic properties of the hopping operators .
more precisely , we show that the particles obey statistics @xmath38 if @xmath39 for any three hopping operators @xmath40 , where @xmath41 are ( distinct ) neighbors of @xmath42 ( ordered in the clockwise direction in the case of @xmath0 dimensions ) . the orientation convention in @xmath0 dimensions is necessary for the anyonic case . the simplest examples where we can apply this formula are the cases of noninteracting hard - core bosons or fermions .
in these cases the @xmath10 particle hamiltonian can be written as @xmath43 where the @xmath44 s are the boson or fermion annihilation operators .
the hopping operators are just @xmath45 .
a little algebra confirms that the boson and fermion hopping operators do indeed satisfy with @xmath46 and @xmath16 , respectively .
we now give a general derivation of the formula .
we begin with the state @xmath47 which contains particles at sites @xmath48 and other particles far away .
imagine that we exchange the two particles at @xmath48 using an appropriate product of hopping operators .
one expression for the final swapped state is @xmath49 here , we ve grouped the hopping operators into 5 terms .
each of these terms moves a particle along some path on the lattice .
the cumulative effect of all 5 terms is to exchange the particles ( see fig .
[ stat]a ) .
similarly , we can construct a path which does nt swap the two particles .
one expression for the final unswapped state is @xmath50 once again , we ve grouped the hopping operators into 5 terms . in this case , the cumulative effect of these terms is simply to move the two particles along independent loops without any swapping taking place ( see fig . [
stat]b ) . .
the numbers label the order in which the paths are traversed .
( b ) the path of the particles in the unswapped state @xmath51 .
notice that the two states only differ in the ordering of the paths .
however , in one case the particles switch places , while in the other , they do not.,width=240 ] intuitively , we expect that the phase difference between the swapped and unswapped states is related to the statistical phase of the particles .
in fact , the special way we ve constructed our states allows us to make a much stronger statement .
notice that the two states involve the same product of hopping operators .
the only difference is the order of these operators .
this means that , in the swapped and unswapped states , the two particles trace out the same total path in configuration space .
this is important because it implies that any phase which comes from gauge fields or other berry phases contributes equally to the swapped and unswapped states .
the phase difference between the swapped and unswapped states is therefore exactly equal to the statistical phase of the particles .
this intuitive argument can be made rigorous using the discussion in the previous section .
indeed , it s not hard to see that the two space - time paths traced out by the swapped and unswapped states are exactly the same locally - they only differ by a reordering .
furthermore , the phase difference between the swapped and unswapped states can be written as a phase difference between two expressions of the form ( [ prod ] ) .
thus , taking the phase difference between the swapped and unswapped states is completely equivalent to the procedure derived in the previous section .
we now compute this phase difference .
we use the assumed algebraic relation @xmath52 applying this relation to the swapped state and reordering the hopping operators using locality , we find @xmath53 the phase difference is @xmath38 , so the particles obey statistics @xmath38 , as claimed . a similar formula can be derived for the relative statistics of distinguishable particles .
this formula , together with its derivation , is given in the appendix .
in this section we consider the properties of a lattice spin system with fermionic or anyonic excitations .
we argue that the excitations are always created in pairs , and the creation operator for a pair of particles has a string - like structure , with the new particles located at the ends .
one interpretation of this is that fermions never appear alone - they always come with some kind of gauge field .
we now make these statements more precise .
suppose we have a lattice spin system with fermionic excitations ( the anyonic case is completely analogous ) .
the total hilbert space of the lattice spin model is a direct product of local hilbert spaces @xmath54 , each associated with an individual lattice site @xmath55 .
we expect that the total hilbert space contains a low energy subspace which is spanned by @xmath10 fermion states .
the corresponding low energy effective hamiltonian is given by @xmath56 where @xmath8 is a projection operator onto the @xmath10 fermion subspace .
typically , @xmath57 can be written as a sum @xmath58 where the @xmath33 are hopping operators .
we expect that the @xmath33 are local in the underlying spin degrees of freedom .
that is , they act only within the local hilbert spaces near sites @xmath42 , @xmath59 .
are _ local at low energies _ : the @xmath33 can be written as @xmath60 where @xmath61 is local and commutes with the projection operator @xmath8 .
] we would like to point out that the lattice on which the fermion hops ( labeled @xmath42 ) may not be the same as the lattice formed by the local hilbert spaces ( labeled @xmath55 ) .
now , consider the string - like product of hopping operators @xmath62 this operator destroys a fermion at site @xmath63 and creates one at site @xmath42 : @xmath64 where @xmath65 , @xmath66 are low - energy states with fermions at @xmath63 , @xmath42 , ( and possibly other fermions far away ) . furthermore
, this operator clearly has a string - like structure : it is made up of a string - like product of operators , each of each is local in the underlying spin degrees of freedom .
the only issue is that this string operator might be trivial .
that is , @xmath67 might act trivially on all the intermediate spins at @xmath68 , and only have a nontrivial effect near @xmath42 and @xmath63 . in that case
, @xmath67 is nt really a string at all .
this is a legitimate concern , since the string operator typically is trivial in the case of _ bosonic _ quasi - particles .
consider , for example , a lattice of noninteracting spins in a magnetic field : @xmath69 .
the ground state has @xmath70 for all @xmath42 .
the excitations are obtained by flipping one spin : @xmath71 for some @xmath59 .
if we perturb the hamiltonian by a term @xmath72 , these excitations acquire dynamics .
the corresponding hopping operators are then @xmath73 and the string operator is @xmath74 the operator @xmath67 clearly creates a particle at position @xmath42 and destroys one at position @xmath63 . however , as we see above , it is trivial and has no string - like structure .
the particle creation and annihilation operators are completely local .
our claim is that this can never happen in the case of emerging fermions .
that is , the string operator @xmath67 can never be written as a product @xmath75 where @xmath76 , @xmath77 are operators which act only within the local hilbert spaces near sites @xmath42 , @xmath63 .
physically , this means that the fermion creation and annihilation operators are never local in the underlying bosonic degrees of freedom .
instead , they naturally appear in pairs , with a string - like structure connecting them . can not be written as a product @xmath78 where @xmath79 is the projection operator onto the @xmath10 fermion subspace with no particles at any of the intermediate sites @xmath80 , and @xmath81 are local operators commuting with @xmath8 .
this statement is slightly stronger then the one made in the text , but the same proof applies . ] _ proof : _ we wish to use the algebraic formula ( [ swap ] ) from the previous section .
that formula relied on the locality of the hopping operators ( [ loc ] ) . here
we would like to use a weaker locality condition : @xmath35 = 0 \label{weakloc}\ ] ] if @xmath42 , @xmath59 , _ are far _ from @xmath82 , @xmath83 .
one can make a similar argument using this weaker assumption .
one arrives at a slightly weaker algebraic relation : @xmath84 if @xmath42 , @xmath83 , @xmath82 are sufficiently far from each other . since @xmath85
, we can rewrite this as @xmath86 condition ( [ fermi ] ) is a direct consequence of the fermionic nature of the quasi - particles .
we will now show that it implies that the strings are nontrivial .
indeed , if the @xmath67 could be written as @xmath87 , then @xmath88 but the @xmath89 s and @xmath90 s are local operators , so they commute with each other when they are well separated .
we can therefore rearrange the operators in these two equations to obtain @xmath91 this directly contradicts the fermion condition ( [ fermi ] ) .
we conclude that the string operator is always nontrivial .
@xmath92 the presence of this nontrivial string indicates that fermions always appear together with some kind of gauge field .
one way to see this is to consider a closed loop of hopping operators @xmath93 .
this string can be interpreted as a wilson loop operator , since its phase is precisely the accumulated phase of the particle when it traverses a loop .
the fact that it is nontrivial ( that is , not equal to the identity operator ) means that the particle is coupled to a nontrivial gauge field .
in this section , we present an exactly soluble lattice spin model with fermionic excitations .
the exactly soluble model provides a concrete realization of the string picture discussed above .
the emerging fermions turn out to be coupled to a @xmath94 gauge field . in this spin-1/2 system , proposed by kitaev@xcite , the spins live on the _ links _ of a square lattice ( fig .
[ slatt3 ] ) .
the hamiltonian is @xmath95 here @xmath96 labels the links , @xmath97 labels the sites , and @xmath98 labels the plaquettes of the square lattice .
also , @xmath99 denotes the loop connecting the four spins adjacent to site @xmath97 , while @xmath100 denotes the loop connecting the four spins adjacent to plaquette @xmath98 ( see fig .
[ slatt3 ] ) . and
@xmath100 are drawn as dotted diamonds , and are labeled with @xmath101 s or @xmath1 s according to whether the corresponding term in the hamiltonian involves @xmath102 s or @xmath103 s .
, width=2 ] this model is exactly soluble since all the terms in the hamiltonian commute with each other .
the ground state satisfies @xmath104 for all sites @xmath55 , and @xmath105 for all plaquettes @xmath98 .
there are two types of ( localized ) excited states .
we can have a site where @xmath106 or we can have a plaquette where @xmath107 we call the first type of excitation a `` charge '' and the second type of excitation a `` flux . '' static charge and flux configurations are exact eigenstates of the above hamiltonian .
thus , the charge and flux quasi - particles have no dynamics .
this lack of dynamics is a special feature of the above model .
however , we are interested in the properties of a _ generic _ hamiltonian in the same quantum phase .
thus , we need to perturb the system , and analyze the resulting dynamics .
the simplest nontrivial perturbation is @xmath108 it s not hard to see that the first term allows the fluxes to hop from plaquettes to adjacent plaquettes , while the second term allows the charges to hop from sites to neighboring sites .
we now calculate the statistics of the fluxes . to do this ,
we restrict our hamiltonian to the low energy subspace with @xmath10 fluxes and zero charges . within this subspace
, our hamiltonian reduces to @xmath109 to make contact with our previous formalism , we write this as @xmath110 where the sum is taken over adjacent plaquettes , @xmath111 , @xmath112 is defined by @xmath113 , and @xmath96 the link joining @xmath98 and @xmath114 . to calculate the statistics , we need to compare @xmath115 with @xmath116 .
well , it s obvious from the definition that all the hopping operators @xmath112 commute with one another .
therefore , @xmath117 we conclude that the fluxes are bosons . in the same way
, one can show that the charges are also bosons .
next , we consider the bound state of a flux and a charge .
that is , we consider excitations with a flux through some plaquette @xmath98 and also a charge at one of the sites @xmath97 adjacent to @xmath98 . these bound states are not actually stable for the above hamiltonian ( [ h1 ] ) - the charge and flux will separate from one another over time .
however , one can imagine modifying the hamiltonian so that charges and fluxes prefer to be adjacent to each other . in this case
, the bound state is a true quasi - particle .
suppose we ve made such a modification .
we can then consider the statistics of the bound state .
if we restrict our hamiltonian ( [ h1 ] ) to the low energy subspace with @xmath10 bound states , then our hamiltonian reduces to @xmath118 to understand the effect of these terms , imagine we have a bound state with a flux at @xmath98 and a charge at @xmath97 .
it s not hard to see that the first term allows the flux to hop to the two neighboring plaquettes which are also adjacent to @xmath97 .
similarly , the second term allows the charge to hop to the two neighboring sites , adjacent to @xmath98 .
all other hopping destroys the bound state and is therefore forbidden for energetic reasons .
thus , the perturbation gives rise to @xmath119 types of hopping operators - @xmath0 corresponding to fluxes and @xmath0 corresponding to charges .
( see fig .
[ bstate ] ) . formally , we can write our low - energy hamiltonian as @xmath120 where the hopping operators are defined by @xmath121 depending on whether @xmath122 differ by a flux hop or a charge hop .
in the first case , @xmath96 is defined to be the link joining @xmath98 and @xmath114 , while in the second case , @xmath96 is the link joining @xmath97 and @xmath123 . .
the flux hops are denoted by dotted arrows , and have hopping operator @xmath102 .
notice that each charge hopping operator anti - commutes with a corresponding flux hopping operator , but everything else commutes .
, width=1 ] to calculate the statistics we need to compare a product of the form @xmath124 with the product @xmath125 , where the @xmath126 are hopping operators involving a single bound state at position @xmath127 .
( we could write out these expressions precisely , but it s messy and not very enlightening )
. now , as we discussed above , there are four different ways that a bound state at @xmath127 can hop - two charge hops , and two flux hops .
each of the flux hopping operators can be paired with a corresponding charge operator which involves the same link @xmath96 .
it s easy to see that each of the flux operators anti - commutes with the corresponding charge operator ( since one involves a @xmath102 , while the other involves a @xmath103 ) .
however , everything else commutes .
( see fig .
[ bstate ] ) . with these facts in mind , we can compare @xmath124 with @xmath125 . by our discussion above ,
any set of @xmath1 hopping operators involving @xmath127 must contain exactly @xmath0 which anti - commute .
thus , exactly @xmath0 of @xmath128 anti - commute .
this implies that @xmath129 we conclude that the bound states are fermions . of course , this result is not that surprising once we notice that the charges and fluxes have relative statistics @xmath130 .
we can easily check that the charges and fluxes have relative statistics @xmath16 .
normally , we would just use the formula ( [ rel ] ) derived in the appendix .
however , this relation assumes that the two types of particles both hop on the same lattice . in our case ,
the charges hop on the square lattice , while the fluxes hop on the dual lattice .
it s not hard to see that in our case the formula ( [ rel ] ) needs to be modified to @xmath131 where @xmath132 are neighboring sites , @xmath111 are neighboring plaquettes , and the links connecting @xmath132 , and @xmath111 are the same . from our previous calculations we know that @xmath133 , and @xmath134 .
therefore , @xmath112 , @xmath135 anti - commute , and the relative statistics are @xmath130 .
we now see why the two particles were called `` charges '' and `` fluxes '' - they have the same statistics and relative statistics as @xmath94 charges and fluxes .
it turns out that this connection with @xmath94 gauge theory extends beyond the low energy regime - in fact , all the way to the lattice scale .
one can show that the kitaev model is exactly equivalent to standard @xmath94 gauge theory coupled to a @xmath94 higgs field .
we argued earlier that whenever fermions or anyons occur in a bosonic system , they are always created in pairs , and the pair creation operator has a string - like structure .
the above exactly soluble model ( [ hkit ] ) provides a good example of this phenomenon .
we begin with the charges .
we can construct the string operators associated with these particles by taking products of their hopping operators along some path @xmath136 on the lattice .
we find @xmath137 where @xmath138 are the links along the path ( see fig . [ string ] ) .
notice that @xmath139 is nontrivial - it acts non - trivially on the spin degrees of freedom along the string @xmath8 .
it is also a pair creation operator : if we apply @xmath139 to the ground state , the resulting state is an exact eigenstate with two charges - one located at each endpoint of @xmath8 .
one way to see this is to notice that @xmath139 commutes with everything in the hamiltonian except @xmath140 and @xmath141 .
the string anti - commutes with these two operators , so when we apply it to the ground state , we get a state with two charges located at @xmath142 .
the case of the flux quasi - particles is very similar .
we take the product of the flux hopping operators along some path @xmath143 on the dual lattice .
we find @xmath144 where @xmath138 are the links along the path ( see fig . [ string ] ) . just as before , one can show that the string @xmath145 is a creation operator which creates two fluxes at the endpoints of @xmath146 .
finally , consider the case of the bound state of the charge and the flux .
the hopping operator for the bound state is a combination of the charge and flux hopping operators , so the string turns out to be a combination of the charge and flux strings .
let @xmath147 be a path in bound state configuration space .
then the associated bound state string operator is @xmath148 where @xmath149 are the links along the path and @xmath150 or @xmath1 depending on whether @xmath151 differ by a flux hop or a charge hop , respectively ( see fig .
[ string ] ) .
once again , one can show that the string @xmath152 is a creation operator for a pair of bound states located at the ends of @xmath153 .
in each of these examples , the creation operator for fermions or anyons has an extended string - like structure .
it is important to note that the position of this string is completely unobservable .
that is , in each case , the excited state @xmath154 is independent of the position of @xmath8 : it only depends on the position of the _ endpoints _ of @xmath8 .
in two dimensional systems , one can create a fermion by binding a @xmath2 vortex to a @xmath2 charge .
this is how we obtained the fermion in the above spin-1/2 model on the square lattice .
since both the @xmath2 vortex and @xmath2 charge appear as the ends of open strings , the fermions also appear as the ends of strings .
however , in three dimensions , we can not change a boson into a fermion by attaching a @xmath155-flux .
thus one may wonder if fermions still appear as the ends of strings in ( 3 + 1)d . in this section ,
we study an exactly soluble spin-3/2 model on a cubic lattice .
we will show that the creation operator for fermions does indeed have a string - like structure .
this example demonstrates that the string picture for fermions is more general then the flux - charge picture .
our model has four states for each site of a cubic lattice .
thus we call it a spin-3/2 model .
let @xmath156 , @xmath157 , @xmath158 , be @xmath159 hermitian matrices that satisfy @xmath160= & 0,\ \
\hbox{if $ a , b , c , d$ are all different } \nonumber\\ \ga^{ab}\ga^{bc } = & i \ga^{ac } , \
\ a\neq c \nonumber\\ ( \ga^{ab})^2 = & 1\end{aligned}\ ] ] a solution of the above algebra can be constructed by taking pairwise products of majorana fermion operators @xmath161 , @xmath162 : @xmath163 the six majorana fermion operators naturally require a space of dimension @xmath164 , but if we restrict them to the space @xmath165 , we obtain the desired @xmath159 hermitian matrices .
alternatively , a more concrete description of the @xmath156 is given the appendix , where we express the @xmath156 in terms of dirac matrices .
in terms of @xmath166 , the exactly soluble spin-3/2 hamiltonian can be written as @xmath167 where @xmath98 labels all the square plaquettes in the cubic lattice , and the @xmath168 are `` flux '' operators defined by @xmath169 depending on the orientation of the plaquette @xmath170 . just as in the kitaev model,@xcite
all the @xmath168 commute with each other and all the @xmath168 have only two eigenvalues : @xmath171 .
thus , we can solve the hamiltonian by simultaneously diagonalizing all the @xmath168 . if @xmath172 is a common eigenstate of @xmath168 with @xmath173 , @xmath174 , then it is also an energy eigenstate with energy @xmath175 the one subtlety is that the @xmath176 are not all independent .
if @xmath177 is the surface of a unit cube , then we have the operator identity @xmath178 it follows that @xmath179 for all cubes @xmath177 .
this constraint means that the spectrum of our model is identical to a @xmath94 gauge theory on a cubic lattice .
the ground state can be thought of as a state with no flux : @xmath180 for all @xmath98 .
similarly , the elementary excitations are small flux loops where @xmath181 for the four plaquettes @xmath170 adjacent to some link @xmath182
. we can think of these excitations as quasi - particles which live on the _ links _ of the cubic lattice .
we would like to compute the statistics of these excitations .
it is tempting to assume that they are bosons since the model is almost the same as a @xmath94 gauge theory .
however , as we will show , this superficial similarity is misleading : the flux loops are actually fermions .
as in the kitaev model , the quasi - particles are exact eigenstates of our hamiltonian .
thus , they have no dynamics and it is difficult to compute their statistics .
however , this is a special feature of our model . we need to perturb the theory to understand generic states in the same quantum phase .
the simplest perturbation is @xmath183 to compute the statistics , we need to restrict ourselves to the low energy subspace with @xmath10 quasi - particles . in this subspace , @xmath184 reduces to @xmath185 the effect of each term @xmath186 is to allow the quasi - particles to hop between the two links @xmath187 and @xmath188 adjacent to site @xmath189 . here
@xmath190 if @xmath191 , @xmath192 if @xmath193 , .
( see fig .
[ 3dhop ] . )
thus , our hamiltonian can be written in the standard hopping form : @xmath194 where @xmath195 . to calculate the statistics we need to compare a product of the form @xmath124 with the product @xmath125 , where the @xmath126 are hopping operators involving a single link @xmath182 .
( as in the kitaev model , we could write out these expressions explicitly , but it s not very enlightening ) .
note that a quasi - particle on the link @xmath182 can hop to any of the @xmath196 neighboring links , @xmath197 of which are adjacent to site @xmath189 , and @xmath197 of which are adjacent to site @xmath198 ( see fig .
[ 3dhop ] ) . using the algebra ( [ gabalg ] ) or the majorana fermion representation , we find that the @xmath197 hopping operators associated with site @xmath189 all anti - commute with each other and similarly for site @xmath198 . on the other hand , each of the operators associated with @xmath189 commute with each of the operators associated with @xmath198 . can hop onto 10 different links : @xmath0 - @xmath199 .
the hopping @xmath200 is generated by @xmath201 , the hopping @xmath202 by @xmath203 , and the hopping @xmath204 by @xmath205 .
, width=144 ] with these facts in mind , we can now compare @xmath124 with @xmath125 .
there are essentially two cases : either all three of the @xmath126 s are associated with a single site @xmath189 or @xmath198 , or two involve one site and one involves the other . in the first case ,
all the @xmath126 s anti - commute ; in the second case , a pair of the @xmath126 s anti - commute and everything else commutes . in either case
, we have @xmath129 we conclude that the quasi - particles are indeed fermions . next , we construct the associated string operator . taking a product of hopping operators along a path ( of links ) , @xmath206 , gives @xmath207 where @xmath208 is the pair of indices associated with the link @xmath209 .
( see fig.[3dlatt ] . )
notice that the string operator commutes with the hamiltonian , except at the ends .
thus , when we apply @xmath139 to the ground state , the only effect is to flip the signs of @xmath176 for the plaquettes @xmath98 adjacent to the links @xmath210 , @xmath211 .
this means that the open string operator creates a pair of fermions at its two ends . .
, width=249 ]
in this paper we have derived a general relation ( [ swap ] ) between the statistics of a lattice particle and the algebra of its hopping operators .
this relation allows us to analyze emerging fermions in three dimensions .
we have also shown that there is a close connection between strings and fermions .
the statistical algebra ( [ swap ] ) is fundamental to this connection since it allows us to determine the statistics of the ends of strings from the structure of the string operator .
it is interesting to put this string picture of emerging fermions in context .
indeed , it is well - known that gauge theories and strings are closely related - the strings correspond to electric flux lines in the associated gauge theory .
@xcite thus , it appears that the fundamental concepts of fermi statistics , gauge theory , and strings are all connected .
they are all just different aspects of a kind of order - topological order - in bosonic lattice systems .
this research is supported by nsf grant no .
dmr0123156 and by nsf - mrsec grant no . dmr0213282 .
in this section , we derive an algebraic formula , analogous to ( [ swap ] ) , for the relative statistics of distinguishable particles .
we begin with a hilbert space which describes two types of hard - core particles hopping on a 2d lattice . for concreteness ,
say that there are @xmath3 particles of type @xmath101 and @xmath10 particles of type @xmath0 .
the states can be labeled by listing the positions of the two types of particles : @xmath212 .
a typical hamiltonian for this system is of the form @xmath213 where @xmath214 are hopping operators for the two types of particles .
we wish to calculate the relative statistics of the two types of particles . as before ,
the statistics are related to the simple algebraic properties of the hopping operators .
specifically , we will show that the particles have relative statistics @xmath215 if @xmath216 here , @xmath217 are ( distinct ) neighbors of @xmath218 oriented in the clockwise direction . .
the dotted line is the path of particle @xmath101 , and the solid line is the path of particle @xmath0 .
the numbers label the order in which the paths are traversed .
( b ) the path of the particles in the unwound state @xmath219 .
notice that the two states only differ in the ordering of the paths .
however , in one case the particles wind around each other , while in the other , they do not.,width=240 ] we begin with the state @xmath220 which contains a type @xmath101 particle at @xmath42 , a type @xmath0 particle at @xmath59 , and other particles which are far away .
imagine that we wind particle @xmath101 around particle @xmath0 using an appropriate product of hopping operators .
one expression for the final wound state is : @xmath221 here , we ve grouped the hopping operators into 6 terms .
the cumulative effect of these terms is to wind one particle around the other ( see fig . [ relstat]a ) .
similarly , we can construct a state where the two particles do nt wind around each other ( see fig .
[ relstat]b ) : @xmath222 as before , the two states involve the same total path in configuration space .
thus , the phase difference between the two states is precisely the relative statistics of the particles .
this can be made rigorous using an argument similar to the exchange statistics case .
we now calculate this phase difference .
we use the assumed algebraic relation @xmath223 applying this relation to the wound state , and reordering the hopping operators using locality , we find @xmath224 this establishes the desired result : the particles have relative statistics @xmath215 . a good consistency check for this formula can be obtained by considering the relative statistics of _ two particles of the same type_. in this case , we expect the angle for the relative statistics to be exactly twice the angle for the exchange statistics : @xmath225 one way to see this is that exchanging the particles twice in the same direction is topologically equivalent to winding one particle around the other .
this result , which has a topological character to it , can be derived algebraically from our two formulas .
we start with the expression @xmath226 . applying the exchange statistics formula ,
we find : @xmath227 applying the formula again gives @xmath228 combining these two equations gives @xmath229 comparing this with the relative statistics formula , we see that @xmath230 , as claimed .
in the following , we represent @xmath156 s in terms of the dirac matrices . note that implies that @xmath231 thus @xmath232 , @xmath233 , @xmath234 , and @xmath235 satisfy the algebra of dirac matrices . introducing four dirac matrices
@xmath236 we can write @xmath237 similarly , @xmath238 , @xmath239 , @xmath240 , and @xmath241 satisfy the algebra of dirac matrices .
we can express those operators as @xmath242 where @xmath243 . finally ,
for @xmath244 , we have @xmath245 in this way , we can express all the @xmath156 , @xmath157 , @xmath158 in terms of dirac matrices .
|
we investigate the general properties of lattice spin models with emerging fermionic excitations .
we argue that fermions always come in pairs and their creation operator always has a string - like structure with the newly created particles appearing at the endpoints of the string .
the physical implication of this structure is that the fermions always couple to a nontrivial gauge field .
we present exactly soluble examples of this phenomenon in @xmath0 and @xmath1 dimensions .
our analysis is based on an algebraic formula that relates the statistics of a lattice particle to the properties of its hopping operators .
this approach has the advantage that it works in any number of dimensions - unlike the flux - binding picture developed in fqh theory .
|
coronal mass ejections ( cmes ) are large scale eruptions of magnetized plasma from the solar atmosphere into the interplanetary space .
it is generally accepted that their energy source is derived from the free energy contained in sheared or twisted magnetic fields @xcite .
many cme models have been developed and despite the differences in the underlying physics of the eruption , all models at some point involve a magnetic flux rope . from this viewpoint ,
the models can be split into those which require the flux rope to exist prior to the eruption , and those in which the flux rope is formed as a result of topological changes in the course of the eruption . in the first category
the rope is fundamental to the cme initiation process .
these models include ideal mhd instabilities of a flux rope , the force imbalance between a rope and its overlying arcade - like field , and the catastrophe of the rope - arcade configuration as a whole . in the second category
, it is proposed that a sheared magnetic arcade becomes unstable to upward expansion . in the course of the expansion , a vertical current sheet forms in the bottom part of the arcade .
magnetic reconnection in the sheet transforms the inner part of the arcade into a growing flux rope starting early in the expansion process and supporting the expansion in a positive feedback .
obviously , determining whether the pre - eruption magnetic field topology involves a flux rope is key to understanding the physics of cme initiation .
how a flux rope can be formed in the corona prior to an eruption is an open question as well .
it may emerge from the convection zone or be formed in situ by an arcade - to - rope topology transformation .
numerical simulations of flux emergence indicate that the process essentially stops as the magnetic axis of the rope hits the photosphere , producing a sheared arcade in the corona @xcite .
moreover , pre - cme formation of large - scale flux ropes between active regions and in the quiet sun can not occur by emergence .
the theoretical predictions for the evolution of an emerged arcade range from further continuous shearing , keeping the arcade topology until an eruption occurs after 12 weeks , to the immediate formation of a coronal flux rope , which can find a stable equilibrium or erupt readily , depending on the relative strength and orientation of emerging and preexisting field .
the observations indicate that active regions typically produce eruptions several days after their emergence is largely complete .
this is consistent with both , gradual arcade shearing and gradual arcade - to - rope topology transformation .
dispersal and diffusion of the photospheric flux concentrations and flows converging toward the polarity inversion line ( pil ) of the photospheric field are main drivers of such quasi - static coronal evolution , which often involves flux cancellation in the photosphere along the pil .
the observational evidence for the existence of a flux rope topology is most compelling for the evolved stage of a cme and gets weaker as earlier phases are considered .
interplanetary data reveal that at 1 au many cmes are well described by a flux rope model and coronal images often suggest a flux rope topology for erupting filaments and cme cores . some erupting filaments exhibit a rotation of their axis about the direction of ascent ; this results from a conversion of twist to writhe and implies a flux rope topology . on the other hand ,
it is unclear whether filaments in their stable equilibrium prior to eruption are contained in fields of flux rope or arcade topology @xcite as it is not yet possible to directly measure the three - dimensional structure of the coronal field .
measurements at the photospheric boundary have supported the flux rope topology , however @xcite .
the pre - eruption flux rope formation processes mentioned above involve reconnection at or above the pil , which , in the case of gradual evolution , is presumably weak and intermittent and in principle not different from the ubiquitous small - scale reconnection events in the solar atmosphere .
therefore , to our knowledge , it has not yet been possible to infer the formation of a flux rope from signatures of reconnection during the quasi - static evolution prior to eruptions .
support was obtained from the observation of an s shaped emission source at low coronal temperatures , interpreted as cooling plasma on newly reconnected field lines within a flux rope .
events of flux convergence and cancellation at a large - scale pil presumably represent the clearest indication of the formation process , but are no proof .
another indication , though not a proof either , is the `` necking '' of coronal cavities in the final stages of their evolution toward eruption . here
we present an investigation of coronal reconnection signatures that demonstrate the formation of a flux rope prior to the onset of a cme .
we study a sigmoidal active region ( s shaped in x - rays ) from the quasi - static evolution , appearing as a long - lived sigmoid , to its eruption , which produced two topologically different stages of a transient sigmoid .
such sources are a signature of locally enhanced dissipation most likely reconnection , and their occurrence is correlated with cmes .
transient sigmoids are formed in quasi - separatrix layers ( qsls ) , which represent the interface between two topologically distinct flux systems ; in the case of sigmoids a flux rope and its surrounding arcade .
this may also be true for long - lived sigmoids ; alternatively , these may just represent a collection of highly sheared coronal loops , i.e. , an arcade .
the qsl in a flux rope - arcade system can be of two different topologies .
it can either intersect itself under the rope along a line of magnetic x - type topology , or extend down to the bottom of the atmosphere .
the latter situation is expected for the emergence of a sub - photospheric flux rope s full cross section as suggested in @xcite . which of the two topologies results from the process of arcade - to - rope transformation , depends upon how high in the atmosphere the reconnection proceeds and whether all reconnected flux under the x line submerges below the photosphere , dragging the x line downward .
both are open questions .
we will focus on the topology without an x line , since this is relevant to the data investigated here . in the absence of an x - type topology under the rope , the qsl touches the pil tangent to the photosphere , with its field lines pointing in the inverse direction ( from the negative to the positive side of the pil ) .
sections of the pil having this property appear as bald patches in h@xmath0 images , so the corresponding qsl has become known as bald - patch separatrix surface , bpss .
the field lines passing through bald patches must wrap around the flux rope to connect to their end points on the other side of the pil .
hence , they are sigmoidal in shape , being forward ( reverse ) s for right - handed ( left - handed ) chirality . in the process of flux rope formation by arcade - to - rope transformation or by rope emergence ,
the bald - patch field lines have the highest likelihood of all the field lines in the configuration to be illuminated in x - rays because they pass through the volume of reconnection , at least temporarily .
the hot plasma resulting from reconnection expands in the corona along these field lines .
due to inertial line tying in the photosphere , the bottom of the bpss must stay largely immobile at coronal timescales .
however , the top part of the flux rope can tear apart from the bottom section .
this makes the bpss topology relevant for so - called partial eruptions which produce a cme but leave behind filament material in the bottom part of the source volume . in the event of eruption , the bottom part of the bpss survives and is likely to be illuminated further as the rising upper part of the flux rope tears off , causing a steepening of the currents in the bpss and , correspondingly , further dissipation .
however , a partial flux rope eruption also forms a vertical current sheet above the pil and inside the original rope , i.e. , topologically distinct from the bpss .
once formed , it becomes the place of strongest energy release , and the flare loops emerging in its downward reconnection outflow are likely to become the brightest x - ray source . from the above it
is clear that a sigmoid can be expected to occur as a signature of flux rope formation in the corona and , reversely , that the observation of a bpss sigmoid implies a flux rope topology .
this sigmoid must be a continuous s crossing the pil three times , different from a sheared arcade whose loops generally cross the pil only once .
when the flux rope experiences a partial eruption , the middle part of the bpss sigmoid remains stationary , while a new , brighter x - ray source appears above it , which evolves into the post - eruption arcade .
noaa active region ( ar ) 8005 hosted a foward s sigmoid and produced a cme on 1996 december 19 which was associated with a _ goes _ c2.3 class x - ray flare at 15:21 ut ( figure [ fig : lightcurves ] ) .
we study the coronal and photospheric evolution leading to the event using , respectively , half and quarter resolution almg and al.1 filter images recorded by the soft x - ray telescope ( sxt ) onboard _ yohkoh _ and magnetograms obtained by the michelson doppler imager ( mdi ) onboard _ soho _ . ]
the active region was in its decay phase as it rotated over the limb , already having dispersed magnetic polarities and no sunspots . in the three days leading up to the event , the x - ray loops showed a gradual evolution to larger extent , whilst the appearance of a sheared arcade with an overall s shape was maintained ( figure [ fig : sigmoid_evolution ] ) .
this appearance results from the j shape of the individual loops , produced as a direct consequence of the accumulation of longitudinal flux ( pointing along the pil ) in the centre of active regions in the course of their long - term , largely diffusive evolution .
the dominance of this flux gives all arcade field lines in the middle of the region a straight leg and permits them to pass over the pil only at the edges of the region , producing the j shape .
beginning at about 00 ut on december 19 , the sigmoid narrowed . at 01:22 ut a weak
, non - eruptive flare occurred near its northwestern end .
after the flare had faded , the sigmoid consisted of two sets of sharper j shaped loops ( threads ) with shortened straight legs . by 10:37 ut the first j shaped threads had merged into a continuous s , and by the exposure at 13:57 ut the evolution of all visible threads into a continuous s shaped sigmoid was complete ( figure [ fig : sigmoid_evolution ] ) . as justified below , we regard this to be a bpss sigmoid .
it had a dip in intensity in its central portion .
the sxt images taken during the main rise of the flare ( 15:3115:48 ut ) show that the continuous s shaped sigmoid survived the onset of the eruption .
its eastern elbow remained visible under the growing arcade of flare loops for at least two hours , well into the flare decay phase .
moreover , the sigmoid showed no signatures of motion .
its elbows turned around less and shrank somewhat in the course of the flare , but the central part remained stationary .
the rise of the soft x - ray flux originated in a new source located in the core of the preexisting sigmoid , overlying its dip .
the new source was closely aligned to the preexisting sigmoid s central section , but with strands seen branching off , so initially it formed a new sigmoid with clearly different end points .
it evolved into the post - eruption arcade , which suggests that the new sigmoid was composed of the first , highly sheared flare loops , formed in the downward reconnection outflow of a vertical current sheet .
the evolution of the new sigmoid into the post - eruption arcade was described in greater detail by , who also concluded that it was a structure different from the preexisting sigmoid .
h@xmath0 and helium 10830 data reveal the presence of a filament situated along the pil , whose central and southeastern section survived the eruption , as seen in the available images at 15:52 , 16:57 , and 20:33 ut . the evolution of the photospheric flux was dominated by ongoing dispersal of the main flux concentrations , with no significant emergence of new flux , throughout the disk passage of the active region ( see the animated mdi data accompanying figure [ fig : cancellation ] ) .
this included several episodes in which patches of strong flux detached from the main polarities , approached the pil , and cancelled .
one of these occurred at the northwestern end of the sigmoid between about december 18 , 21 ut and december 19 , 08 ut and was obviously associated with the small preceding flare .
another , stronger cancellation episode occurred between about december 19 , 03 ut and december 20 , 11 ut in a narrow region under the middle of the sigmoid , where the two j s merged into the continuous s ( figure [ fig : cancellation ] ) .
it was preceded by cancellations of weak flux elements in the same area since about december 18 , 20 ut .
these data indicate that the changes in the sigmoid were not causally related to the preceding flare , but were driven primarily by the cancellation episode shown in figure [ fig : cancellation ] .
co - aligned mdi and sxt data on december 1819 show that each of the two sets of point symmetric j shaped loops made one crossing of the pil and had clearly discernible end points at either side of it .
when the continuous sigmoid formed , it made three pil crossings ; twice by the sigmoid elbows and once as its central part crossed the pil in the inverse direction ( figure [ fig : mdi_overlays ] ) .
overall , the pil ran at about @xmath1 inclination to the meridional lines , parallel to the central part of the sigmoid .
this is a well known characteristic of sigmoids which so far has prevented the proof of an inverse pil crossing .
however , at the point where the major flux cancellation on december 1920 occurred and the continuous sigmoid formed , the pil ran locally nearly east - west , at a substantial inclination to the sigmoid at that point .
the observations of the sigmoid and the partial eruption of the active region in a cme and flare conform very clearly to the expected signatures of a flux rope with a bald - patch separatrix surface .
all key features theoretically predicted were found in the data .
these include ( 1 ) the formation of a continuous s shaped transient sigmoid from a diffuse , arcade - like long - lived sigmoid , ( 2 ) the transient sigmoid s triple pil crossing , ( 3 ) the co - location of the inverse pil crossing , the final merging into a continuous s trace , and the area where cancelling strong flux patches converged , ( 4 ) the association with a _ partial _ filament eruption , ( 5 ) the transient sigmoid s survival through the flare peak , and ( 6 ) the stationary nature of its central part during the eruption .
the combination of features 2 , 5 , and 6 can not be attained by field lines in an erupting sheared arcade .
partial eruptions and the triple pil crossing of some field lines within an arcade can be individually arranged in simulations by applying specific photospheric shear profiles . however , a partial eruption that leaves just the s shaped field lines within the sheared volume in place and essentially stationary is impossible ; the sheared flux would always participate in the expansion to some degree .
moreover , the mdi data do not yield any indication of the required shear flow . consequently , the findings above are conclusive evidence for the formation of a coronal flux rope .
the existence of a bpss in a long - lived , but eventually erupting , sigmoid was conjectured in another recent case study , based primarily on the j shape of the sigmoid threads , which by itself is not conclusive , however ( see above ) . moreover , only the features 1 and 6 were demonstrated , and no information was given as to whether the transformation from the double - j to the single - s appearance of the sigmoid occurred before or in the course of the eruption .
the formation of the bpss sigmoid in the active region considered here began over five hours , and was complete over 80 minutes , prior to the onset of the flare .
a linear backward extrapolation of the cme height - time data yields an onset time 10 minutes before flare onset . since the cme was rather fast ( with a projected velocity of 469 kms@xmath2 in spite of its origin near disk centre ) and the associated flare was impulsive
, it is very unlikely that the cme had a gradual initial evolution with a significantly earlier onset time .
consequently , the data of the sigmoid and eruption imply that a coronal flux rope existed in the active region at least one hour , likely over five hours , _ prior _ to the onset of the cme .
the data support the gradual arcade - to - rope transformation as the mechanism of flux rope formation because the evolution of the flux was dominated by dispersal and cancellation , not by emergence which had occurred several days earlier .
the transition to the bpss topology , as seen in soft x - rays , was a process of at least 14 hours in duration .
however , the straight legs of the j s did not reconnect with each other immediately , and some of their threads kept a slight separation , albeit decreasing , up to the sxt exposure at 13:10 ut .
the question arises of why a continuous s trace was nt visible throughout the transition .
we relate this behaviour to the dominance of the longitudinal flux in the environment of the pil , mentioned above , which is characteristic of old flux concentrations and clearly indicated for the present active region by the formation of a filament and by the double - j shape of the long - lived sigmoid .
such flux is nearly unidirectional and can hardly reconnect inside when perturbed ; essentially it represents a barrier to the approaching flux patches .
two effects are expected to result in the corona from the observed localised perturbation .
their relative importance depends on the angle between the longitudinal and approaching flux , which controls the strength of reconnection .
the approaching flux can slide under the longitudinal flux ( which is rooted remote from the perturbed area ) , lifting it gradually , as often observed for filaments prior to an eruption .
the approaching flux will also reconnect in the interface with the longitudinal flux , illuminating narrow j shaped field line bundles in the interface .
as the approach and reconnection progress , the field lines lit up lie closer to the pil , have one footpoint in the approaching flux , i.e. , nearer to the merging point of the sigmoid , and connect to their remote footpoint in an arc at increasing distance ( because they were brought in from a more remote location and have to encircle a growing flux rope ) .
these consequences of reconnection are all clearly seen in the sxt data ( figure [ fig : sigmoid_evolution ] ) . when the approaching flux eventually reaches direct contact with opposite flux at the pil beneath existing longitudinal flux ( as indicated by the continued presence of the filament ) , the shortened straight legs of the j s merge into a continuous s by reconnecting with each other near their end points .
the flux they were temporarily rooted in submerges in the cancellation event , leaving a flux rope with bpss topology .
we acknowledge the use of data provided by the global high resolution h@xmath0 network , the _
yohkoh_/sxt , _ soho_/mdi , and _ goes _ instruments , and by the cme catalog held at nasa s cdaw data center .
this work was supported by the royal society , stfc , nsf grant atm 0518218 , and nasa grant nnh06ad58i .
amari , t. , luciani , j. f. , aly , j. j. , mikic , z. , & linker , j. 2003a , , 585 , 1073 amari , t. , luciani , j. f. , aly , j. j. , mikic , z. , & linker , j. 2003b , , 595 , 1231 antiochos , s. k. , dahlburg , r. b. , & klimchuk , j. a. 1994 , , 420 , l41 antiochos , s. k. , devore , c. r. , & klimchuk , j. a. 1999 , , 510 , 485 fan , y.2001 , , 544 , 111 forbes , t. g. 2000 , , 105 , 23153 forbes , t. g. , & isenberg , p. a. 1991 , , 373 , 294 gibson , s. e. , & fan , y. 2006 , , 637 , l65 gibson , s. e. , fan , y. , trk , t. , & kliem , b. 2006b , , 124 , 131 gibson , s. e. , foster , d. , burkepile , j. , de toma , g. , & stanger , a. 2006a , , 641 , 590 green , l. m. , kliem , b. , trk , t. , van driel - gesztelyi , l. , & attrill , g. d. r.2007 , , 246 , 365 jian , l. , russell , c. t. , luhmann , j. g. , & skoug , r. m. 2006 , , 239 , 393 kliem , b. , & trk , t. 2006 , phys .
, 96 , 255002 lites , b. w. 2005 , , 622 , 1275 low , b. c. 1996 , , 167 , 217 lynch , b. j. , antiochos , s. k. , devore , c. r. , luhmann , j. g. , & zurbuchen , t. h. 2008 , , 683 , 1192 mackay , d. h. , & van ballegooijen , a. a. 2006 , , 641 , 577 manchester , w. iv , gombosi , t. , dezeeuw , d. , & fan , y. 2004 , , 610 , 588 martin , s. f. 1998 , , 182 , 107 mckenzie , d. e. , & canfield , r. c. 2008 , , 481 , l65 moore , r. l. , sterling , a. c. , hudson , h. s. , & lemen , j. r. 2001 , , 552 , 833 trk , t. , & kliem , b. 2005 , , 630 , l97 tsuneta , s. , et al .
1991 , , 136 , 37 tripathi , d. , kliem , b. , mason , h. e. , young , p.
r. , & green , l. m. 2009 , , 698 , l27 van ballegooijen , a. a. , & mackay , d. h. 2007 , , 659 , 1713 van ballegooijen , a. a. , & martens , p. c. h. 1989 , , 343 , 971
|
we analyse the evolution of a sigmoidal ( s shaped ) active region toward eruption , which includes a coronal mass ejection ( cme ) but leaves part of the filament in place .
the x - ray sigmoid is found to trace out three different magnetic topologies in succession : a highly sheared arcade of coronal loops in its long - lived phase , a bald - patch separatrix surface ( bpss ) in the hours before the cme , and the first flare loops in its major transient intensity enhancement .
the coronal evolution is driven by photospheric changes which involve the convergence and cancellation of flux elements under the sigmoid and filament .
the data yield unambiguous evidence for the existence of a bpss , and hence a flux rope , in the corona prior to the onset of the cme .
|
since our spacetime may ultimately prove to be discrete , field theories on discrete spacetimes have developed an interest in and of themselves @xcite .
in particular , field theories on two - dimensional discrete spacetimes can be interpreted as string theories with a discrete string worldsheet and a continuous target space .
exact solutions for such theories can often be obtained even after quantization ( see , for example , @xcite ) .
yet another approach to discretizing spacetime and constructing a theory of gravity based on stochastic properties of random lattices is advocated by r. sorkin @xcite .
investigation of field theories on discrete spacetime might therefore contribute to our understanding of discrete quantum picture of the universe .
we shall focus on the connection between the discrete - spacetime and the continuous - spacetime versions of the same field theory .
this work was motivated by the ( somewhat `` mysterious '' ) fact noted in ref .
@xcite that a particular discrete version of the two - dimensional wave equation gives _ exact _ solutions of the underlying continuous equation .
in other words , the exact solution of the free scalar field theory coincides , on the lattice points , with the solution of the lattice version of the same theory .
the main intent of this paper is to understand the origin of the exact equivalence of continuous field theories and their discrete counterparts , as well as to find more general classes of ( classical ) field theories that are exactly solved by properly chosen discretizations .
the origin of the exact solution property of discretized equations turns out to be the discrete conformal symmetry of the models @xcite .
the exact solution property of the massless bosonic model has a direct application in numerical simulations of strings ( which was the original context of ref .
a quantized version of the discrete conformal symmetry may prove useful in string theory .
the paper is organized as follows .
we first present the exact solution property of the discretized two - dimensional wave equation . then , in sec .
[ secdci ] , an analysis of the discrete equations leads us to the concept of discrete conformal invariance ( dci ) , and we give general conditions for a given discrete field equation to possess dci . in sec .
[ secgen ] we present a general description of discrete models with dci , and show that they all yield exact solutions for the corresponding ( first and/or second - order ) differential equations ; these solutions are always algebraically factorized .
further examples of equations with dci based on lie groups and semi - groups are presented in sec .
[ examples ] .
conclusions follow in sec .
[ secconcl ] .
the two - dimensional wave equation came out of numerical modeling of cosmic strings @xcite .
the evolution equations for the target space variables @xmath0 are @xmath1 with gauge conditions @xmath2 here , the subscripts @xmath3 and @xmath4 denote differentiation with respect to world - sheet coordinates .
the string world - sheet can be discretized to a rectangular lattice with coordinates @xmath5 and steps @xmath6 @xmath7 . a common second - order discretization of ( [ waveequ ] )
would be @xmath8 however , a special choice of discretization steps , @xmath9 , was adopted in @xcite , and it was noted that the resulting discrete equations are not only simpler , @xmath10 @xmath11 ^2=0 , \label{dwegaugefix}\ ] ] but in addition exactly solve the continuous equation ( [ waveequ ] ) .
namely , one can easily check that the general solution of ( [ waveequ ] ) , @xmath12 ( where @xmath13 and @xmath14 are vector functions of one argument , which are determined from the appropriate boundary conditions ) exactly satisfies the discretized equation ( [ dwaveequ ] ) .
the gauge conditions ( [ dwegaugefix ] ) are then reduced to constraints on @xmath15 , @xmath16 ^2=0 .
\label{dwesolconstr}\ ] ] to try to understand why this happens , let us examine the discretization ( [ dwaveequ ] ) in more detail . expanding ( [ dwaveequ ] ) in powers of the discretization step @xmath17 , we notice that the discrete equation ( [ dwaveequ ] ) , on solutions of @xmath18 , is satisfied to all orders in @xmath19 : @xmath20 since @xmath21 approximation to all orders means exact solution in the following sense .
one can show that @xmath22 where @xmath13 and @xmath14 are now arbitrary vector functions of one discrete argument , is in fact the general solution of ( [ dwaveequ ] ) .
given a boundary value problem for the continuous equation ( [ waveequ ] ) , we can restrict the boundary conditions to the discrete boundary points and find the discrete functions @xmath15 .
then , from the form of the solution ( [ dwesol ] ) it is clear that @xmath23 will coincide , on lattice points , with the solution of the original boundary value problem .
this is what we call the exact solution property of the discretization ( [ dwaveequ ] ) . like in the continuous case ,
the lines @xmath24 are characteristics of eq .
( [ dwaveequ ] ) .
it is easily checked that the gauge conditions ( [ wegaugefix ] ) are compatible with ( [ dwaveequ ] ) , in the sense that if the discretized gauge conditions ( [ dwegaugefix ] ) hold at some point @xmath5 , then the evolution equation ( [ dwaveequ ] ) guarantees that they hold everywhere along the characteristic that intersects the point @xmath5 .
however , the exact solutions of the equations ( [ waveequ])@xmath25([wegaugefix ] ) do not necessarily satisfy ( [ dwegaugefix ] ) for any discretization step @xmath17 , because the continuous constraint ( [ wegaugefix ] ) imposed on the derivative of @xmath15 is non - linear and does not integrate to a discrete constraint ( [ dwesolconstr ] ) . in this sense , the discrete constraints ( [ dwegaugefix ] ) provide only an approximate solution of ( [ wegaugefix ] ) .
we have shown that a particular discretization ( [ dwaveequ ] ) of the wave equation exactly solves the corresponding continuum equation , regardless of the magnitude of the lattice step @xmath17 .
the remaining part of this work is intended to explain and explore this unexpected phenomenon .
the `` mysterious '' cancellation of all higher - order terms in ( [ dweseries ] ) is due to conformal invariance , as will be explored in the next two sections .
since the exact solution property is independent of the discretization step @xmath19 , we are motivated to explore some kind of scale invariance in the model .
we first notice that eq .
( [ dwaveequ ] ) does not mix the values of the field on `` even '' and `` odd '' sublattices , each sublattice consisting of points with @xmath26 even or odd .
therefore we shall further consider only one of these two sublattices , which is itself a square lattice in the lightcone coordinates @xmath27 . in fig . 1 we drew four adjacent squares , each square corresponding to a group of four adjacent points on the lightcone lattice with the field values @xmath28 related by ( [ dwaveequ ] ) , for instance @xmath29 by adding together the corresponding equations of motion for all four squares in fig .
1 , it is straightforward to show that the same equation actually holds for points @xmath30 , @xmath31 , @xmath32 , and @xmath33 that lie at the vertices of a larger @xmath34 square , i.e. @xmath35 in a similar fashion we find that the discrete equation ( [ dwaveequ ] ) remains valid if we replace the discretization step @xmath17 by its multiple @xmath36 , which means removing all but @xmath37-th points from the lattice , and effectively stretching the lattice scale by a factor of @xmath37
. we could call this fact _ scale invariance _ of the discrete equations .
in fact , an even more general kind of invariance holds : namely , eq .
( [ dwaveequ ] ) retains its form when applied to the four vertices of any rectangle on the lightcone lattice ( i.e. with sides parallel to the null directions ; such a rectangle would have its sides parallel to the @xmath38 axes ) .
one can prove this by noticing that , by adding eq .
( [ dwaveequ ] ) applied to points @xmath39 and @xmath40 , one obtains @xmath41 which is the same equation applied to points @xmath42 which lie at vertices of a rectangle rather than a square .
the same procedure shows that eq .
( [ dwaveequ ] ) applies to points @xmath43 as well . after deriving ( [ dwaveequ ] ) for the `` elementary '' @xmath44 rectangles ,
it is clear that , for any given rectangle , one only needs to add together the relation ( [ dwaveequ ] ) for all squares inside it to obtain the same relation for the vertices of the rectangle .
the fact that eq .
( [ dwaveequ ] ) applies to vertices of any lattice rectangle means that if we remove all points lying on any number of lines @xmath45 and @xmath46 from the lattice , the discrete equations will still apply to the remaining points . in lightcone
coordinates , such a transformation amounts to replacing the ( discrete ) coordinates @xmath47 by functions of themselves : @xmath48 here , the functions @xmath49 are arbitrary monotonically increasing discrete - valued functions of one discrete argument .
the monotonicity of these functions is necessary to preserve causal relations on the spacetime .
we immediately notice a similarity between ( [ dctrans ] ) and conformal transformations in a two - dimensional pseudo - euclidean space .
as is well known , the wave equation ( [ waveequ ] ) allows arbitrary conformal transformations of the world - sheet coordinates , @xmath50 this is the most general form of the coordinate transformation that preserves the pseudo - euclidean metric @xmath51 up to a conformal factor
. transformations ( [ dctrans ] ) are obviously the discrete analog of conformal transformations ( [ twodcoordchange ] ) on the lightcone lattice . we have , therefore , found that the discrete wave equation ( [ dwaveequ ] ) is invariant under the discrete conformal transformations ( [ dctrans ] ) .
we shall refer to this as the _ discrete conformal invariance _ ( dci ) of the discrete wave equation .
the natural question is then whether there exist other equations with the property of dci , and whether such equations also deliver exact solutions of their continuous limits . in the remaining sections
, we will answer this question in the positive , establishing a relation between the exact solution property and dci .
note that the constraint equations ( [ dwegaugefix ] ) do not preserve their form under discrete conformal transformations .
for instance , if the constraints hold for some step size @xmath17 , @xmath52
^2=0,\quad \left [ x^\mu \left ( \tau , \sigma + \delta \right ) -x^\mu \left ( \tau -\delta , \sigma + 2\delta \right ) \right ] ^2=0,\ ] ] it does not follow in general that they also hold for step size @xmath53 : @xmath54 ^2\neq 0.\ ] ] accordingly , as we have seen in the previous section , the discrete constraint equations do not exactly solve their continuous counterparts .
to try to generalize eq .
( [ dwaveequ ] ) , we write the generic discrete evolution equation on a square lightcone lattice as @xmath55 , \label{devol}\ ] ] where @xmath56 is an unknown function . as we have found in the previous section , the property of dci will be satisfied if ( [ devol ] ) is invariant under the `` elementary '' conformal transformations , that is , if the relation ( [ devol ] ) is valid when applied to the vertices of all @xmath57 lattice rectangles .
this requirement can be written as two functional conditions on @xmath58 : [ dciconds ] @xmath59 here , @xmath30 , @xmath31 , @xmath32 , and @xmath33 are arbitrary field values which may be scalar or vector ( or even belong to a non - linear manifold of a lie group , as in our examples below ) , and @xmath56 is a similarly valued function . before we try to solve these conditions for @xmath56 , we would like to show that for any function @xmath58 satisfying ( [ dciconds ] ) , the discrete evolution equation ( [ devol ] ) exactly solves its continuous limit .
the continuous limit of ( [ devol ] ) is obtained by expanding it in powers of @xmath17 , for instance @xmath60 and with the assumption that @xmath61 , which is natural if we suppose that a constant function @xmath62 must be a solution of ( [ devol ] ) , we obtain the following equations corresponding to first and second powers of @xmath17 : @xmath63 @xmath64 here , @xmath65 and @xmath66 are derivatives of @xmath56 with respect to its three numbered arguments ( which are in general vector arguments , but we suppressed the indices in the above equations ) .
one can easily verify that for @xmath67 , these equations give the usual wave equation ( [ waveequ ] ) .
the derivatives @xmath65 and @xmath68 are constrained by eqs .
( [ dciconds ] ) .
for example , the first derivatives satisfy @xmath69 where @xmath65 are understood as linear operators on the tangent target space .
it follows that @xmath65 are projection operators ( i.e. operators @xmath70 that obey @xmath71 ) with eigenvalues @xmath72 and @xmath73 only .
now we shall show that the equations ( [ contequ1])@xmath25([contequ2 ] ) are exactly solved by ( [ devol ] ) if the dci conditions ( [ dciconds ] ) hold .
denote the exact solution of the continuous equations ( [ contequ1])@xmath74([contequ2 ] ) by @xmath75 , and the solution of the discrete equation ( [ devol ] ) by @xmath76 . here ,
@xmath3 and @xmath4 are discrete lattice coordinates , and we assume that @xmath77 on the lines @xmath78 ( fig . 2 ) . since eqs .
( [ contequ1])@xmath25([contequ2 ] ) were obtained from ( [ devol ] ) by expansion in @xmath17 up to third - order terms , @xmath79 satisfies the discrete equation ( [ devol ] ) up to @xmath80 , i.e. @xmath81 in the limit of small @xmath17 .
now we shall scale up the lattice by an arbitrarily chosen factor of @xmath37 .
as we have seen in the previous section , from dci it follows that eq .
( [ devol ] ) applies also to the vertices of an @xmath82 square on the lightcone lattice , and this is a consequence of using eq .
( [ devol ] ) @xmath83 times , once for each `` elementary '' square . at each elementary square
, we can replace the field values @xmath84 by @xmath79 and introduce an error of @xmath85 .
an error of @xmath86 in @xmath87 entails an error of not more than @xmath88 in @xmath89 , because the first derivatives of @xmath56 are operators with eigenvalues of @xmath72 and @xmath73 only .
therefore , replacing @xmath84 by @xmath79 in the @xmath90 square entails an error in @xmath91 of not more than @xmath92 : @xmath93 however , we can also apply ( [ cubeerror ] ) directly to the @xmath82 square to obtain @xmath94 it means that @xmath95 which forces @xmath96 since @xmath97 is a polynomial in @xmath17 .
therefore , the approximation ( [ cubeerror ] ) is in fact exact .
this argument shows that it was in fact only necessary to expand ( [ devol ] ) up to second order in @xmath17 , and all higher - order terms will lead to differential equations which are consequences of ( [ contequ1])@xmath25([contequ2 ] ) , just as we have seen in the case of the wave equation .
it also shows that the differential equations corresponding to given dci field equations ( [ devol ] ) are always second - order or lower .
using the evolution equation ( [ devol ] ) , one can write the exact solution of the continuous equations for boundary conditions on the lightcone @xmath98 . to find @xmath99 at a point @xmath100 within the future lightcone of the origin
, we construct a lattice that has @xmath5 as one of its points , and then apply eq .
( [ devol ] ) to the rectangle @xmath101 and explicitly write @xmath99 through the boundary values : @xmath102 now we try to describe a general class of functions @xmath56 satisfying ( [ dciconds ] ) .
first , we find by combining eqs .
( [ dciconds ] ) that @xmath103 if we denote @xmath104 the above will look like an associative law for a binary operation @xmath105 : @xmath106 the condition @xmath107 looks like a unity law @xmath108 although it does nt follow from ( [ dciconds ] ) that @xmath109 for all @xmath31 .
however , if we suppose that ( [ multi ] ) is actually an operation of group multiplication , with the usual unity law @xmath110 and the inverse operation @xmath111 , then it is shown in the appendix that there exists a reparametrization @xmath112 of the field values such that the operation @xmath105 is written as @xmath113 where by @xmath114 we denote the group multiplication in the group of field values .
note that although eq .
( [ genm ] ) is not symmetric with respect to interchange of @xmath30 and @xmath31 , such interchange is equivalent to a reparametrization @xmath115 ( where @xmath116 is the group inverse of @xmath30 ) .
we arrive at a picture of a field equation of the type ( [ devol ] ) , derived from a group multiplication in an arbitrary lie group .
we shall assume that any needed reparametrization is already effected , and that the evolution is directly given by ( [ devol ] ) with @xmath117 for example , if we consider a vector space as an abelian group with addition of vectors as the operation @xmath114 , we again obtain the formula @xmath118 for the discrete wave equation . in view of eq .
( [ genf ] ) , the exact solution ( [ exactsol ] ) has an algebraically factorized form , as a product of functions of the lightcone coordinates : @xmath119 ( the constant @xmath120 is absorbed by either of the functions ) .
this is a characteristic feature of the dci equations we are concerned with . the continuous limit of the discrete equations
( [ devol])@xmath25([genf])can also be described in terms of an arbitrary lie group @xmath121 as a target space ; it is a wess - zumino - witten ( wzw)-type model @xcite .
if @xmath87 is a field with values in a ( matrix ) group @xmath121 defined on a two - dimensional manifold @xmath122 , the action functional of the wzw model is defined by @xmath123 where @xmath124 is an auxiliary @xmath125-dimensional manifold whose boundary is @xmath122 , and @xmath126 is the matrix trace operation . for a specific choice @xmath127 , the equations of motion in the lightcone
coordinates become @xmath128 with the general solution @xmath129 with @xmath130 being arbitrary @xmath121-valued functions .
it is immediately seen that the solution ( [ wzwsol ] ) is the same as ( [ esolgroup ] ) , if we ( naturally ) choose @xmath114 to be the group multiplication in @xmath121 .
in this section we present some examples to illustrate the constructions of sec .
[ secgen ] and to give a physical interpretation of the field theories arising from them .
as was noted in the previous section , we can obtain the discrete wave equation by using the general formula ( [ genf ] ) on a vector space considered as an additive group of vectors .
since all commutative lie groups are locally isomorphic to a vector space , we should take a non - commutative group to find a less trivial example .
the simplest non - commutative lie group has two parameters and can be realized by matrices of the form @xmath131 the composition law of the group is @xmath132 or , written more compactly , @xmath133 the inverse element is given by @xmath134 .
a calculation shows that the continuous limit of the discrete evolution equation ( [ genf ] ) with the composition law ( [ twodmc ] ) is ( in lightcone coordinates @xmath47 ) [ twodme ] @xmath135 note that the non - commutativity of the group leads to asymmetry with respect to the spatial reflection ( i.e. interchange of @xmath136 and @xmath137 ) , but at the same time spatial reflection is equivalent to reparametrization @xmath138 @xmath139 which corresponds to taking the inverse matrix to ( [ matrixtwod ] ) . the exact solution of ( [ twodme ] ) is obtained from the general formula ( [ exactsol ] ) : @xmath140 with arbitrary functions @xmath141 and @xmath142 . the equations ( [ twodme ] ) were obtained from a two - dimensional group and describe a pair of coupled scalar fields .
more generally , one may start with an arbitrary @xmath37-dimensional non - commutative group @xmath121 and construct the corresponding discrete and continuous equations describing @xmath37 coupled scalar fields .
the group structure will then be reflected in the coupling of the fields : for example , if the group @xmath121 has a commutative subgroup @xmath143 , then the corresponding parameters will satisfy a free wave equation .
this can be easily seen from the example above : the elements of the form @xmath144 form a commutative subgroup , and the corresponding equation ( [ twodmealpha ] ) for @xmath145 is a wave equation .
so far , we have been dealing with scalar fields .
now we shall consider the possibility of dci equations describing fermions .
the 2-dimensional massless dirac equation for the two - component fermionic field @xmath146 is @xmath147 where the corresponding dirac matrices satisfy the usual relations of a clifford algebra @xmath148 and can be chosen as @xmath149 here , the metric @xmath150 in the lightcone coordinates @xmath151 is @xmath152 , @xmath153 . with the choice ( [ twoddiracgammas ] ) , eq .
( [ twoddirac ] ) becomes @xmath154 where @xmath155 are left- and right - moving components of the field @xmath156 .
the corresponding lattice equations are [ twoddiracd ] @xmath157 their solution is @xmath158 which is also an exact solution of the continuous equations ( [ twoddiraclc ] ) .
( here , @xmath159 are functions of discrete argument determined by boundary conditions . ) since we again discovered a case of an exact solution , we naturally try to write the lattice equations ( [ twoddiracd ] ) in the form ( [ devol])@xmath25([genf ] ) .
this is possible if we define the multiplication operation @xmath114 by @xmath160 ( note how the left- and right - handed components of the field propagate to the left and to the right of the multiplication sign . )
such a multiplication operation is associative , but does not allow a unity element and is not invertible , which would make it impossible to write @xmath161 as in ( [ genf ] ) . however , this operation has the property that @xmath162 regardless of the value of @xmath163 , and therefore we can disregard @xmath120 in ( [ genf ] ) .
a matrix representation of this multiplication operation can be defined by @xmath164 since the derivations of sec .
[ secgen ] only use the associativity of the multiplication operation @xmath114 , all our considerations apply also to cases where this operation does not have an inverse , such as in the case of _ semigroups _ @xcite
. we shall be interested in a specific example of the semigroup structure represented in ( [ matrixsemigroup ] ) , where @xmath165 and @xmath166 can be , in general , multi - component fields ; we shall refer to such a structure as a `` fermionic semigroup '' .
heuristically , a lie group generates bosons and a fermionic semigroup generates fermions in dci field theories .
the direct product of a group and a semigroup would result in a theory describing uncoupled bosons and fermions .
an example of a model containing coupled bosons and fermions can be obtained from a semigroup built as a _ semi - direct product _ of a group and a fermionic semigroup .
a semi - direct ( or `` twisted '' ) product of two semi - groups @xmath167 and @xmath121 can be defined if @xmath167 _ acts _ on @xmath121 , i.e. if for each @xmath168 there is a map @xmath169 such that @xmath170 and @xmath171 ( here , the multiplication denotes the respective semi - group operation in @xmath167 or @xmath121 , where appropriate . )
the semi - direct product of @xmath167 and @xmath121 is the set of pairs @xmath172 with the multiplication defined by @xmath173 one can easily check that this operation is associative .
of course , a group is also a semigroup , and the semi - direct product construction can be applied to two groups or to a group and a semigroup as well .
the existence of an associative binary operation on the target space is really all we need to build a dci field theory .
we can obtain a generic theory of this kind containing both bosons and fermions by taking the fermionic semigroup @xmath167 and some lie group @xmath121 on which @xmath167 acts .
such pairs @xmath174 can be constructed for arbitrary dimensions of @xmath167 and @xmath175 .
we shall , to illustrate this construction , couple the two previous examples and arrive to a model with two interacting bosons and three fermions . to do this , we take @xmath167 to be the simplest fermionic semigroup of ( [ matrixsemigroup ] ) . as @xmath121
we choose a group like one represented by ( [ matrixtwod ] ) , but with two parameters @xmath176 : @xmath177 to couple bosons with fermions , we need to introduce some action of @xmath167 on @xmath178 . for example , we can multiply the column of parameters @xmath176 of the group @xmath121 by the right - moving part of the matrix @xmath179 , @xmath180 which defines the action of @xmath181 on @xmath182 as @xmath183 one can check that the conditions ( [ maphom])@xmath25([hommap ] ) hold for such an action .
the multiplication law of the resulting five - parametric semigroup is @xmath184 with a matrix realization @xmath185 the multiplication in this semigroup does not allow an inverse operation .
nevertheless , the problem with @xmath120 in the general formula ( [ genf ] ) is circumvented because we have chosen the twisting of the semigroup and the group in such a way as to make the product @xmath186 independent of the @xmath187 components of @xmath32 .
the continuous limit equations in this model are @xmath188 as can be seen from the above , @xmath155 and @xmath189 are fermions , while @xmath190 and @xmath191 are bosons , @xmath191 is coupled to @xmath145 and @xmath189 is coupled to @xmath166 and @xmath191 .
( here , we formally refer to the fields with first - order equations of motion as `` fermionic '' . )
similar models can be constructed for a larger number of coupled bosonic and fermionic fields .
note that the field @xmath189 which was a boson in the @xmath192 model , became a coupled fermion after we added the fermionic sector .
we have explored the phenomenon of the exact solution of continuous equations by their discretizations .
we formulated the property of discrete conformal invariance ( dci ) , and showed that any system of lattice equations possessing dci delivers exact solutions of its continuous limit . in this sense , the conformal invariance is the cause of the exact solution property ; the continuous limit equations must also be conformally invariant ( although not all conformally invariant equations are exactly solved by any discretizations ) .
we found a class of lattice equations , based on lie group target space , with the property of dci ; their continuous limit corresponds to a theory of wzw bosons .
we also found a more general class of theories based on semigroups describing scalar fields and fermions , which can be in general nonlinearly coupled to each other .
in all these models , solutions to boundary value problems can be written explicitly ( see eq .
( [ exactsol ] ) ) using the multiplication law of the group or semigroup at hand . expressed in this fashion through the boundary conditions on a lightcone
, the solutions are always algebraically factorized . in case of the wave equation ( [ waveequ ] ) with gauge constraints ( [ wegaugefix ] )
, it was found that the discretized constraint equations ( [ dwegaugefix ] ) are also exactly solved by certain solutions of the discretized wave equation , however they are not conformally invariant and , correspondingly , the exact solutions of the equations ( [ waveequ])@xmath25([wegaugefix ] ) do not necessarily satisfy ( [ dwegaugefix ] ) for any discretization step @xmath17 .
the `` compatibility '' of the discretized equations ( [ dwaveequ ] ) and ( [ dwegaugefix ] ) is perhaps due to the simple algebraic form of the solutions ( [ dwesol ] ) .
the author is grateful to alex vilenkin for suggesting the problem and for comments on the manuscript , and to itzhak bars , arvind borde , oleg gleizer , leonid positselsky , and washington taylor for helpful and inspiring discussions .
here we show that if the binary operation @xmath193 on target space @xmath194 is not only associative but is actually a group multiplication in some group @xmath121 , then the target space can be reparametrized so that the operation @xmath105 becomes , in terms of the group multiplication @xmath114 , @xmath195 by assumption , for each @xmath196 there is a one - to - one map @xmath197 from the target space to the group @xmath121 such that @xmath198 the first condition of ( [ dciconds ] ) can be written as @xmath199 or , if we take @xmath200 of both parts , @xmath201 if we now denote @xmath202 and @xmath203 , where @xmath163 and @xmath204 are elements of @xmath121 , this relation becomes @xmath205 this shows that @xmath206 is almost a group homomorphism , and we can make it one if we define @xmath207 where @xmath208 is a map @xmath209 .
after this ( [ rel2 ] ) becomes @xmath210 for all @xmath211 .
now , obviously @xmath212 is the identity map , and @xmath213 for all @xmath214 .
this means that @xmath215 can be expressed as @xmath216 ^{-1},\ ] ] where @xmath217 is an appropriately chosen homomorphism of @xmath175 , and @xmath218 ^{-1}$ ] is the inverse of the homomorphism @xmath219 .
for example , we could choose an arbitrary element @xmath220 and define a map @xmath221hom@xmath121 by @xmath222 now , if we modify the function @xmath223 by a @xmath224 transformation : @xmath225 ^{-1}g_b\left ( a\right ) , \ ] ] then we obtain @xmath226 which similarly means that @xmath227 is of the form @xmath228 ^{-1},\ ] ] where @xmath229 is an appropriately chosen 1-to-1 map , and @xmath230 ^{-1}$ ] is the group inverse of @xmath231 .
discrete analogs of continuous symmetries have been applied to obtain exact solutions for stochastic lattice systems ; see , for example , m. henkel and g. schtz , int . j. mod
b*8 * , 3487 ( 1994 ) .
the author is grateful to malte henkel for communication on the topic .
|
a discretized massless wave equation in two dimensions , on an appropriately chosen square lattice , exactly reproduces the solutions of the corresponding continuous equations .
we show that the reason for this exact solution property is the discrete analog of conformal invariance present in the model , and find more general field theories on a two - dimensional lattice that exactly solve their continuous limit equations .
these theories describe in general non - linearly coupled bosonic and fermionic fields and are similar to the wess - zumino - witten model .
|
previous studies of strings in backgrounds with compact extra - dimensions have led to an increased understanding of the formation and evolution of string loops with non - trivial windings in the internal space .
both generic scenarios and compactification schemes specific to type iib string theory have been investigated @xcite-@xcite .
+ in the case of static string loops , it has been shown that the presence of a lifting potential in the compact space traps the windings - giving rise to loops with step - like winding configurations referred to in the literature as cosmic necklaces @xcite-@xcite . from a four - dimensional perspective
the windings appear as a series of monopoles or beads " connected by ordinary sections of string .
superficially these resemble the standard string - monopole networks found in field - theoretic models @xcite but their behaviour is , in many ways , fundamentally different . in stark contrast to previous predictions based on field theory defects
, it was found that the gravitational collapse of necklaces formed in the class of backgrounds defined by the klebanov - strassler ( ks ) geometry @xcite leads to the formation of primordial black holes ( pbhs ) @xcite during a window in the early universe , followed by the formation of potential dark matter ( dm ) relics in the scaling regime @xcite .
this is almost the complete reverse of the standard predictions for string - monopole networks @xcite .
+ the root cause of this difference appears to be a time - dependent bead mass in the necklace model , as opposed to the constant bead mass of true monopoles connected by strings .
this arises from the time - dependence of the lifting potential and is a somewhat unexpected result .
initial investigations of string necklaces assumed the existence of a constant potential and hence a constant bead mass , though these were based solely on generic arguments @xcite-@xcite .
the work presented in @xcite represents the first explicit realisation of necklace formation in string theory , in which it was shown that these assumptions must be modified , at least for the class of backgrounds considered .
this raises two interesting possibilities : either the formation of necklace - like objects is possible only in string theory , or there exist previously unknown solutions in field theory models which are equivalent to the objects described by strings with step - like windings .
+ at present there is no known field theory analogue of necklaces formed from extra - dimensional windings . in particular , there is no known way to produce string - monopole networks with time varying bead mass .
the question of whether a corresponding field theory model exists is important because , if the formation of necklace - type objects is specific to models with compact extra dimensions , their predicted impact on observable parameters could be used to obtain experimental evidence in favour of string theory , or at least in favour of higher - dimensional models .
the aim of this paper is to investigate the possibility of such a configuration .
we seek to establish a relation between the topological winding number , @xmath6 , of a defect string ( in four spacetime dimensions ) and the physical winding number , @xmath7 , of an @xmath8/@xmath9-string in a higher - dimensional background .
we also seek to establish a field - theoretic model in which
bead"/pinch " formation occurs dynamically , by analogy with the wound - string case @xcite-@xcite .
+ for simplicity we consider the most basic field - theoretic model of local string formation - the abelian - higgs model .
a non - cylindrical , or
pinched " string solution - which represents generalisation of the well known nielsen - olesen vortex @xcite - naturally arises if the usual abelian - higgs action is modified to include coordinate - dependent couplings in both the scalar and vector fields ( i.e. @xmath10 and @xmath11 in our gauge choice , where @xmath12 represents the position along the length of the string ) .
this appears to mimic the behaviour of the type of necklaces discussed above .
+ we propose the pinched string solution as a model corresponding to wound - string necklaces , and argue that a time - dependent bead mass may be obtained .
the main basis for the proposed correspondence is an analysis of the four - dimensional effective tension of the wound - strings , together with the periodically varying tension of the pinched string , which take the same form for appropriate ansatz choices . following the correspondence between string theory and field theory parameters suggested by this comparison
, we propose a geometric interpretation of the field - theoretic terms , such as gauge flux and topological winding number , in the string picture .
although the introduction of @xmath12-dependence in the original field couplings may appear unnatural , it has a very natural interpretation in the string picture due to the relation between the three - sphere radius @xmath13 and the string coupling @xmath14 ( @xmath15 ) .
+ as the interchange between vortices and anti - vortices in the field - theoretic model necessarily involves the consideration of sub - planckian scales , we present a hypothetical model for discretising the planck - scale structure of the vortex . though such discretisation is necessary to prevent divergences in the euler - lagrange equations , it is not intended to be literal .
whatever the exact nature of the physical limit imposed by the planck scale , it is likely to determine a limit on the process of measurement rather than implying the outright discretisation of spacetime .
we adopt this method only as an approximation to an ( as yet ) unknown theory of the quantum structure of fields on the smallest scales , and present an argument for its validity ( as a toy model ) based on matching boundary solutions of planck - sized regions to well known solutions valid on scales @xmath16 . once again , we find that the localisation of the ( classical ) field - theoretic string core on scales @xmath17 admits a very natural interpretation in terms of the quantum constraints on the corresponding wound @xmath8/@xmath9-strings . + the structure of this paper is as follows : in section 2 we present a brief overview of the abelian - higgs model , including the general ( covariant ) form of the euler - lagrange equations . for completeness , and to allow easy comparison with later calculations of the @xmath12-dependent tension of the pinched string @xmath18 , an outline of the calculation of the constant string tension @xmath19 for the cylindrically symmetric ansatz is reviewed in section 3 . in section 4
we introduce a non - cylindrically symmetric ansatz based on a specific discretisation scheme in the planck - scale region of the vortex core ( in which vorticity itself may no longer be defined ) , and introduce the @xmath12-dependent couplings @xmath10 and @xmath11 . from this
we derive the specific form of the equations of motion ( eom ) to verify that the pinched string configuration is a valid solution . under a set of reasonable assumptions ,
these are found to reduce to simplified forms which ( for fixed @xmath12 ) are structurally equivalent to the eom for the cylindrically symmetric case .
this allows analogous asymptotic and small @xmath20 solutions for the scalar and gauge fields but their behaviour is now , generally , @xmath12-dependent .
we show that the @xmath12-dependent tension depends crucially on the form of a generic , dimensionless , periodic function @xmath21 which varies between zero and one but which , to good approximation , is independent of the physics of the vortices in the planck - scale regions .
we then discuss the effective four - dimensional tension of an @xmath8/@xmath9-string with linear winding ansatz at the tip of the ks throat , and compare with the results for our pinched string . for an appropriate and natural choice of @xmath21
we see that this enables a correspondence between the field theory and the string theory parameters to be inferred .
this section also contains a more detailed analysis of the relation between the scalar and vector fields in the higgs model , and their geometric interpretation in the theory of wound strings .
using the @xmath22 metric convention , and introducing the covariant derivative @xmath23 ( together with its conjugate ) , the covariant form of the abelian - higgs action is @xcite @xmath24 varying this action with respect to the various fields yields the resultant eom , assuming that the gauge field and scalar derivatives vanish at the boundary , @xmath25 with the corresponding maxwell equation now taking the form @xmath26 the @xmath27 current is @xmath28 the static , cylindrically symmetric ansatz then takes the form @xmath29 where @xmath30 are dimensionless variables and @xmath31 are the length scales of the scalar and vector cores , respectively .
these are fixed by the compton wavelengths of the associated scalar and vector bosons and , in the case of the vector core , by the winding number number of the vortex @xmath32 ( which is also equal to the number of units of magnetic flux ) , to be @xcite @xmath33 here @xmath34 and @xmath35 are dimensionless real functions satisfying the conditions @xmath36 and @xmath37 we can then define the scalar and gauge field equations by @xmath38 @xmath39 here we have manipulated the original forms of the eom ( which come from simply substituting the ansatz ( [ eq : scs_ansatz1 ] ) into ( [ eq : eleqn1])-([eq : current1]))by multiplying the original scalar equation by @xmath40 to get ( [ eq : scs_eom1 ] ) and the original vector equation by @xmath41 in order to get ( [ eq : scs_eom2 ] ) .
this allows us to define both the scalar and vector eom in terms of the dimensionless variables @xmath42 and @xmath43 .
+ although multiplying the original form of our equations by powers of @xmath20 may be hazardous in the asymptotic limit , multiplying the vector eom through by a factor of @xmath44 causes no problems for @xmath45 as each term in the equation still goes to zero independently as @xmath46 and @xmath47 .
+ however , defining the parameter @xmath48 we can rewrite the scalar equation as @xmath49 treating the ratio @xmath50 as a numerical constant then allows us to rewrite both the scalar and vector eom in terms of a single dimensionless variable @xmath43 , and adopting the form ( [ eq : scs_eom1_mod ] ) for the scalar eom is equivalent to assuming @xmath51 instead of @xmath52 in the ansatz ( [ eq : scs_ansatz1 ] ) .
this may appear counter - intuitive because we expect the length scale @xmath53 to determine the width of the scalar core ( i.e. the region over which @xmath54 ) .
however , the fact that ( [ eq : scs_eom1 ] ) and ( [ eq : scs_eom1_mod ] ) are algebraically equivalent shows that we may assume either functional form for @xmath55 in our initial ansatz .
both equations have the same approximate solutions in the asymptotic and small @xmath20 limits and it is the value of @xmath53 which characterises the small @xmath20 solution , though both scales @xmath53 and @xmath56 play a role in the asymptotics ( at least when the eom are solved as a coupled pair @xcite ) .
+ we find it convenient to use the form ( [ eq : scs_eom1_mod ] ) instead of ( [ eq : scs_eom1 ] ) for the scalar eom , though
large " and small " @xmath20 for both forms of the equation must still be defined with respect to @xmath53 rather than @xmath56 as for the vector eom ( [ eq : scs_eom2 ] ) . before concluding this section we note the approximate solutions for the functions @xmath55 and @xmath57 within the scalar and vector cores when the eom are solved as an uncoupled pair , i.e. @xmath58 @xmath59 asymptotic solutions to the uncoupled equations
may be found in the usual literatue ( c.f .
@xcite ) and solutions valid for both small and large @xmath20 when the eom are solved as a coupled pair may be found in @xcite .
to lowest order , the small @xmath20 solutions for both functions are not affected by the coupling , so that the expressions above remain approximately valid .
the formula for the tension of an abelian - higgs vortex string is @xmath60 which gives @xmath61 at critical coupling ( @xmath62 ) and @xmath63 when @xmath64 ( for @xmath65 ) @xcite .
this result may be verified explicitly by splitting the integral ( [ eq : mu_1a ] ) into three parts and using the appropriate approximate solutions . within the scalar core region , @xmath66 ,
we may approximate the functions @xmath55 and @xmath57 by @xmath67 and @xmath68 . in the region @xmath69 we assume that @xmath68 and @xmath70 and on scales @xmath71 we may assume that @xmath72 so that @xmath73 and the gauge field contribution effectively cancels the energy density contribution of the scalar field gradient . the overall contribution to the energy density of the covariant gradient term is therefore : @xmath74 turning our attention to the gauge field term
, we see that @xmath75 and the only non - zero component of @xmath76 is @xmath77 so that @xmath78 integrating over the vector core , @xmath79 , for which @xmath68 then gives @xmath80 where we have used the definition @xmath81 .
for @xmath71 we may take @xmath82 so that @xmath83 .
the leading order contribution of the potential term is @xmath84 and , summing all contributions , the total tension becomes @xmath85 although , for @xmath64 , this expression is dominated by the terms proportional to @xmath86 at large winding , at critical coupling @xmath87 we may approximate the tension by @xmath88 note that the numerical factor containing @xmath89 is to some degree arbitrary and depends on the definition of the vector core radius ( here we have employed the definition @xmath90 ) .
the key points are that the leading order contribution at critical coupling is proportional to @xmath32 , which agrees with the expected result , and that at non - critical coupling the tension is logarithmically dependent on @xmath50 .
note that the inclusion of the factor of @xmath91 in the expression for @xmath56 ensures the first of these conditions and this , in turn , ensures the stability of the vortex according to the famous result by bogomolnyi @xcite ( c.f .
also @xcite ) .
+ in the following section we introduce a non - cylindrically symmetric ansatz for the pinched string and solve the resulting eom .
we show that under a set of reasonable physical assumptions , these equations take on analogous forms to those in the more familiar cylindrical case , but with the substitutions @xmath92 , @xmath93 .
in other words , the structure of the vortices remains essentially the same except that the radii of the scalar and vector cores become functions of their position along the string . therefore the calculation of the tension ( for fixed @xmath12 ) remains the same as that outlined above , apart from the introduction of a new radial magnetic flux term @xmath94 ( associated with the @xmath12-dependent variation of @xmath95 ) and of a new @xmath12-derivative component in the scalar gradient which give rise to an additional periodic term .
although the resulting tension @xmath18 is periodic in @xmath12 , it still carries a leading order constant term @xmath96 , derived in an analogous way to that in ( [ eq : convenient_mu1 ] ) , which may be identified with a similar leading order constant term @xmath97 in the expression for the effective four - dimensional tension of a wound @xmath98-string . additionally , by analogy with @xmath99 , we may define the parameter @xmath100 .
taking appropriate ansatz choices for @xmath101 and @xmath102 we have @xmath103 for @xmath104 and the resulting analogue of the logarithmic term @xmath105
( i.e. @xmath106 ) also vanishes .
we now introduce @xmath12-dependent couplings in both the scalar and vector sectors of the abelian - higgs model .
although the couplings are coordinate - dependent , they are not treated as fundamental fields in the same way as @xmath107 and @xmath108 .
we are interested in static , non - cylindrically symmetric solutions to the resulting covariant eom and must therefore modify the ansatz ( [ eq : scs_ansatz1 ] ) to include @xmath12-dependence .
crucially we wish our new solution to represent a string which interpolates between energetically degenerate regions of vortex / anti - vortex solutions to the eom and we anticipate the existence of a planck - sized region in which vorticity itself can not be defined and which separates neighbouring string sections labelled by @xmath0 . to this end
we will also find it necessary to parameterise our new couplings in terms of the winding number @xmath109 where @xmath110 $ ] and @xmath111 $ ] are dimensionless functions .
+ this implies the existence of @xmath12-dependent scalar and vector boson masses ( also paramaterised by @xmath32 ) and hence of @xmath12-dependent effective radii for the scalar and vector cores . in other words ,
a non - cylindrical string can be found from the correspondence @xmath112 by analogy with the cylindrically symmetric case we may define new dimensionless variables using @xmath113 and new @xmath114-dependent functions ( also parameterised by @xmath32 ) @xmath115 and @xmath116 via @xmath117 however we leave the precise form of the @xmath12-dependence unspecified . in order to describe the existence of regions with @xmath118 and @xmath119 winding , as well as the planck - sized region which marks the transition ( and in which @xmath32 becomes undefined ) , we must introduce a step - like function @xmath120 into the field ansatz which takes three distinct values @xmath121 our modified ansatz then takes the form @xmath122 specifically , we can choose to define the function @xmath120 as @xmath123 where @xmath124 and @xmath125 is some scale characterising the length of a section of @xmath0 string ( which we expect to satisfy @xmath126 ) .
we see that @xmath120 admits a representation in terms of a superposition of heaviside step functions @xmath127 , specified over appropriate ranges , such that @xmath128 for future reference we also note that the square of @xmath120 is @xmath129 or equivalently @xmath130 the first derivative of @xmath120 , the square of its first derivative and its second derivative are then @xmath131 \\ & - & \sum_{m=-\infty}^{\infty } \left[\delta\left(z-(2m+1)\delta - \left|n\right| l_p\right ) + \delta\left(z-(2m+1)\delta + \left|n\right| l_p\right)\right ] , \nonumber\end{aligned}\ ] ] @xmath132,\end{aligned}\ ] ] @xmath133 \\ & - & \sum_{m=-\infty}^{\infty } \left[\delta'\left(z-(2m+1)\delta - \left|n\right| l_p\right ) + \delta'\left(z-(2m+1)\delta + \left|n\right| l_p\right)\right ] , \nonumber\end{aligned}\ ] ] where the prime represents differentiation with respect to @xmath12 .
although the mathematical definition of these functions is complicated , the functions themselves are easy to visualise and a plot of @xmath120 is presented in figure 1 . in the range @xmath134 , with @xmath135 and @xmath136.,scaledwidth=60.0% ] note that the planck - sized regions where @xmath137 do not represent genuine discontinuities in the phase of the complex field , or in the rotational direction of the gauge current .
rather they represent regions where both phase @xmath138 and the winding number @xmath139 ( or equivalently the magnetic flux quantum number ) are completely undefined , i.e. we have that , @xmath140 .
\end{aligned}\ ] ] it is not possible ( or meaningful ) to localise the string core on length scales @xmath141 or @xmath142 .
furthermore , we may assume that only a change in topological number of @xmath143 will occur ( on average ) within any planck - sized region , so that it is not possible to define continuous changes in @xmath144 over scales of @xmath145 or @xmath146 , though we contend that this does not necessarily mean that discrete jumps in the topological winding number can not take place over planck - sized distances .
vortex from morphing continuously into an @xmath147 vortex ( @xmath148 ) over length scales @xmath149 as there exists no homomorphism which smoothly maps one state to the other .
discontinuities are also usually considered unphysical .
however on distances @xmath150 there exist no smooth maps at all .
we argue that there is nothing , in principle , to prevent the topological winding number from changing discontinuously as long as it does so in a region where the discontinuity of space is also manifest i.e. over @xmath151 . ]
the change from @xmath152 to @xmath153 winding states must therefore take place over a distance @xmath154 .
the winding number @xmath144 is still undefined over the entire range @xmath155 , but a change in winding number of @xmath156 can not take place over a smaller distance .
it is not , however , meaningful to ascribe a definite change in topological winding number of @xmath143 to a specific planck length within @xmath157 .
+ it therefore seems natural to assume that @xmath144 may change ( on average ) by at most @xmath143 over a single planck length @xmath158 and that the winding number is genuinely undefined at @xmath159 ( where @xmath160 is also undefined ) so that it may be set to zero without loss of generality .
as we traverse the fundamental string core , the absolute value of the winding number changes from @xmath144 down to zero and then back up to @xmath144 which requires a distance of @xmath161 , so that @xmath162 the same bound can also be obtained in a different way : if we assume that is not meaningful for the phase of a complex field @xmath160 to vary continuously over planckian distances , we may estimate the maximum rate of change with respect to the physical angular coordinate @xmath163 at the string core boundary @xmath164 , @xmath165 , via @xmath166 .
but when @xmath167 , we have @xmath168 which recovers the relation above ( [ eq : core_min ] ) .
this is the same as saying that , if the natural unit of phase is one radian , a phase change of @xmath169 can not take place over a distance @xmath170 on the circumference of a circle . )
is seen to be equivalent to the cooper pair ( bogoliubov ) wave - function of a superconducting fluid . considering this in conjunction with the uncertainty principle
then suggests that the minimum scale on which @xmath160 can be defined is @xmath171 .
we then take this to imply that the minimum range over which graduated changes in the scalar field @xmath107 may occur is also @xmath171 .
this applies equally to both the phase and magnitude of the complex field .
additionally the natural unit for the phase @xmath160 is one radian , which together with the considerations above , implies that a phase change of @xmath172 may not take place over a distance @xmath145 and hence gives rise to the condition ( [ eq : core_min ] ) . ] turning now to the eom for the non - cylindrical string , substituting the ansatz ( [ eq : nscs_ansatz1 ] ) into the appropriately modified versions of ( [ eq : eleqn1])-([eq : current2 ] ) gives rise to the following scalar and vector equations : @xmath173 @xmath174 considering ( [ eq : nscs_scalar1 ] ) first , we see that the imaginary part must be set equal to zero independently .
however , the considerations above imply that the phase @xmath160 is effectively undefined not only at @xmath159 but for all @xmath175 over the planck - sized regions in which @xmath176 , @xmath177 $ ] .
we may therefore set @xmath178 in this region without loss of generality , according to ( [ eq : undefined ] ) .
+ in the regions where @xmath179 ( @xmath180 $ ] ) , @xmath160 may be defined consistently but each term in the curly brackets must vanish independently and the imaginary component vanishes for all @xmath12 .
this argument is equivalent to multiplying the entire equation by @xmath120 and setting either @xmath176 or the sum of terms which it multiplies equal to zero in alternate regions .
similar considerations hold for the term proportional to @xmath181 , so that the final form of the scalar eom simplifies to become @xmath182 where we consider only the regions in which @xmath183 .
we may adopt the same strategy when dealing with the vector eom , which shows that all terms in the curly brackets go to zero for all @xmath12 except the term proportional to @xmath184 .
this must be included in the final form of the eom @xmath185 where again we need only consider regions in which @xmath183 . +
this approach is equivalent to treating the planck - scale regions as a black box " for which we have no effective field theory picture .
although we have no explicit expressions for the scalar and vector field functions @xmath186 and @xmath187 , with which to calculate the tension of the string within the ranges @xmath188 $ ] , we will later assume a tension of @xmath189 in these regions to ensure the continuity of @xmath18 .
although this is somewhat unsatisfactory , we will see that both the assumed tension for the planck scale sections and the spatial localisation of the string core to scales @xmath190 have a natural explanation in the string theory picture , which we propose as a justification for the assumptions made here .
+ using our current approach it is impossible to obtain solutions for @xmath191 and @xmath192 in the regions for which @xmath193 , and these remain untreated in our present analysis . clearly if the solutions we obtain in the regions where @xmath183 are to be viewed as physical , the planck - sized regions which connect sections of vortex / anti - vortex string must be dealt with in such a way as to ensure continuity , at least with respect to the string tension @xmath194 .
this problem will be dealt with in the following section , in which @xmath194 is calculated explicitly .
+ although we have yet to specify the precise form of the @xmath12-dependence of @xmath191 and @xmath192 ( in the regions for which @xmath195 ) we may still use our physical intuition to impose appropriate boundary conditions .
it is reasonable to assume that conditions analogous to those imposed on @xmath196 and @xmath197 still hold for any value of @xmath12 , just as they did in the cylindrically symmetric case , so @xmath198 and @xmath199 the only problem that remains is how to specify
large " and small " @xmath20 for fixed @xmath12 . in the cylindrically symmetric case
this was easily solved , since the length scales @xmath53 and @xmath56 determined the radii of the scalar and vector cores , respectively , at every point along the string . in the non - cylindrically symmetric case
we may expect the proposed substitutions ( [ eq : subs1])-([eq : subs3 ] ) to imply the following boundary conditions on the related functions @xmath200 : + @xmath201 + and + @xmath202 we then need only specify the forms of @xmath203 and @xmath204 by introducing an appropriate additional ansatz , before verifying that they solve the eom . to begin we note that @xmath205 and @xmath206\frac{df}{r_{\left|n\right|s}^{eff } } \nonumber\\ & + & \frac{1}{(r_{s|n|}^{eff})^2}(r_{s|n|}^{eff})^2 \left(\frac{dr_{s|n|}^{eff}}{dz}\right)^2 \frac{d^2f}{d(r_{s|n|}^{eff})^2},\end{aligned}\ ] ] and that similar relations exist between @xmath207 and @xmath208 . the scalar eom then becomes @xmath209 \frac{d^2f}{d(r_{s|n|}^{eff})^2 } \nonumber\\ & + & \left[1 + ( r_{s|n|}^{eff})^2\left\{2\left(\frac{dr_{s|n|}^{eff}}{dz}\right)^2 - r_{s|n|}^{eff } \frac{d^2r_{s|n|}^{eff}}{dz^2}\right\}\right ] \frac{1}{r_{s|n|}^{eff } } \frac{df}{dr_{s|n|}^{eff } } \nonumber\\ & + & \frac{|n|^2 f}{(r_{s|n|}^{eff})^2 } - \frac{1}{2}f(f^2 - 1)\left(\frac{r_{s|n|}^{eff}}{r_s}\right).\end{aligned}\ ] ] by analogy with the cylindrically symmetric case , we may then write @xmath210 \frac{d^2f}{d(r_{v|n|}^{eff})^2 } \nonumber\\ & + & \left[1 + ( r_{v|n|}^{eff})^2\left\{2\left(\frac{dr_{v|n|}^{eff}}{dz}\right)^2 - r_{v|n|}^{eff } \frac{d^2r_{v|n|}^{eff}}{dz^2}\right\}\right ] \frac{1}{r_{v|n|}^{eff } } \frac{df}{dr_{v|n|}^{eff } } \nonumber\\ & + & \frac{|n|^2 f}{(r_{v|n|}^{eff})^2 } - \frac{1}{2}\beta_{|n|}^{eff}f(f^2 - 1)\end{aligned}\ ] ] where we have defined @xmath211 hence we see that in the case of the pinched string , the assumptions @xmath212 and @xmath213 are equivalent , in correspondence with the cylindrically symmetric case .
the vector eom becomes @xmath214 \frac{d^2a}{d(r_{v|n|}^{eff})^2 } \nonumber\\ & + & \left[-1 + ( r_{v|n|}^{eff})^2\left\{2\left(\frac{dr_{v|n|}^{eff}}{dz}\right)^2 - r_{v|n|}^{eff } \frac{d^2r_{v|n|}^{eff}}{dz^2}\right\}\right ] \frac{1}{r_{v|n|}^{eff } } \frac{da}{dr_{v|n|}^{eff } } \nonumber\\ & - & |n|f^2(a-1)\left(\frac{r_{v|n|}^{eff}}{r_{v}}\right)^2,\end{aligned}\ ] ] so that in the limit @xmath215 , @xmath216 we recover the usual cylindrically symmetric equations ( [ eq : scs_eom1])/([eq : scs_eom1_mod ] ) and ( [ eq : scs_eom2 ] ) .
+ we now proceed by specifying the conditions we wish @xmath217 and @xmath218 to satisfy .
let us assume that each function @xmath219 , @xmath165 , varies between two values @xmath220 ( in accordance with conditions imposed along with the definitions ( [ eq : subs0 ] ) and ( [ eq : subs1 ] ) , such that @xmath221 and @xmath222.\end{aligned}\ ] ] additionally we impose the following constraint on the derivatives to ensure continuity at the points @xmath223 , @xmath224 which together with ( [ eq : rsneff2 ] ) also implies @xmath225 the most general form for the two functions @xmath219 is @xmath226 where @xmath227 , @xmath165 are periodic functions in @xmath12 , valued between @xmath228 and @xmath229 .
the values of the constants @xmath230 , @xmath231 , @xmath232 and @xmath233 may then be uniquely determined by requiring @xmath234 pass through the three points @xmath235 , @xmath236 and @xmath237 , together with the requirement that the first derivatives ( with respect to @xmath12 ) are zero at the first and last of these points . assuming the functions @xmath238 both have natural period @xmath89
, this gives @xmath239 the variation of the scalar and vector core profiles ( assuming non - critical coupling @xmath240 for a non - cylindrical pinched " string are illustrated in figure 2 using @xmath241 as an example ansatz .
, with @xmath242 , @xmath243 , @xmath244 ( blue curve ) and @xmath245 ( red curve).,scaledwidth=60.0% ] for later convenience we note that for large @xmath164 , satisfying @xmath246 , we can make the approximation @xmath247 so that @xmath248 which becomes effectively independent of @xmath144 ( except for the usual dependence on @xmath91 of the vector core radius ) .
this implies @xmath249 together with @xmath250 in this limit , we may approximate @xmath120 by the function @xmath251 , which is independent of @xmath144 , and defined via @xmath252 whose square is given by @xmath253 for future reference we note that @xmath254,\end{aligned}\ ] ] @xmath255,\end{aligned}\ ] ] and @xmath256 additionally , note that in the limit @xmath257 , we have @xmath258 where @xmath259 if the @xmath12-dependence of @xmath53 and @xmath56 has the same functional form , i.e. if @xmath260 so that @xmath261 then the parameter @xmath262 and becomes effectively independent of both @xmath144 and @xmath12 . at critical coupling , therefore , it too is equal to unity .
+ finally defining the dimensionless variable @xmath263 through @xmath264 the functions @xmath251 , @xmath265 and @xmath266 may be rewritten as @xmath267 , @xmath268 and @xmath269 ( this notation will be useful later ) .
we are now in a position to consider the solutions of ( [ eq : nscs_scalar2])/([eq : nscs_scalar3 ] ) and ( [ eq : nscs_vector2 ] ) : if the terms in the square brackets in each equation satisfy @xmath270 and @xmath271 within the region of the effective string core ( i.e. @xmath272 , or equivalently @xmath273 ) , then the eom take functional forms which are identical ( to within an order of magnitude in the coefficients of the derivative terms ) to those for the cylindrically symmetric ansatz , under the correspondence ( [ eq : subs1])-([eq : subs3 ] ) .
they will therefore admit functionally equivalent solutions under the same correspondence in both the small @xmath20 and asymptotic limits , which may now be defined ( for a given value of @xmath12 ) with respect to @xmath274 for the scalar eom and @xmath275 for the vector eom .
+ let us first consider the constraint ( [ eq : constraint1 ] ) which , using the definitions above , is approximately equivalent to @xmath276 where @xmath277 and a prime denotes differentiation with respect to @xmath278 .
let us further assume that @xmath279 , this assumption may be questioned .
realistically it is necessary to input an initial , smoothly varying , function @xmath238 and to subsequently calculate the initial forces acting on neighbouring string sections due to the interaction of the opposing magnetic fields .
ideally a dynamic theory will be developed which describes the evolution of the string core pinching " under the action of such forces .
we may expect the pinching of the string to to become increasingly localised within the central area of the neighbouring @xmath280 regions , resulting in ever - increasing localisation of the string energy density .
ultimately this would result in the formation of highly localised beads " , and of neighbouring bead - antibead pairs which repel one another .
furthermore each pinch " in the physical radius of the string gives rise to two peaks in the effective tension ( i.e. two beads or antibeads ) associated with the points of maximum absolute gradient @xmath281 .
increasing localisation of the pinching therefore corresponds to the coalescing of neighbouring @xmath282 and @xmath283 pairs as well as the repulsion of neighbouring @xmath284s and @xmath285s .
this could be the analogue of the creation of step - like windings in the formation of cosmic necklaces from smoothly wound strings , which gives rise to similar phenomena @xcite .
if so , a future analysis of the interaction between neighbouring pinches may provide insights into the binding energy between neighbouring bead / antibead pairs in the wound string picture . ] giving @xmath286 since we are interested in the physics at scales @xmath287 , this constraint is essentially equivalent to the stronger condition @xmath288 which appears reasonable on cosmic scales . as long as this condition is satisfied , we may be certain that the terms in the first set of square brackets ( in both scalar and vector eom ) are @xmath289 for all @xmath20 within the effective scalar and vector cores . in the limiting case @xmath290
, this statement still holds . another way to think about this
is to note that the former constraint ( [ eq : oom_arg4 ] ) is equivalent to @xmath291 since @xmath292 \
\forall z$ ] , this effectively reduces to the minimum / maximum conditions @xmath293,\ ] ] or @xmath294 hence the condition @xmath295 implies that ( [ eq : oom_arg4 ] ) is fulfilled for all @xmath175 even at the boundaries of the region @xmath296 ( that is , for all @xmath20 within the effective radius at any value of @xmath12 ) . + it is also possible to identify the distance between neighbouring pinches , @xmath125 , with twice the ( average ) distance between peaks in the periodic tension , i.e. twice the average distance between neighbouring beads or anti - beads .
in this case , by analogy with wound string necklaces , it is possible to identify this with twice the correlation length of the string , @xmath297 , so that the number of strings per hubble volume is @xmath298 and the average distance between neighbouring strings is of order @xmath125 @xcite .
it is therefore likely that condition ( [ eq : oom_arg4n3a ] ) will always be fulfilled .
, may occur , in which case it is doubtful whether the correspondence @xmath299 may be maintained except shortly after the epoch of string network formation .
nonetheless in a fully dynamical theory of pinched string formation / evolution an explicit model of @xmath300 could be obtained , which may also allow the correspondence @xmath301 to hold for all @xmath302 . ] + a similar argument can be used to show that the terms in the second set of square brackets are also of order one ( for the scalar eom ) or of order minus one ( for the vector eom ) for all @xmath12 . imposing the bound ( [ eq : constraint2 ] ) for all values of @xmath12
requires @xmath303 now , assuming @xmath304 , we have @xmath305 which , accounting for the two limiting values of @xmath306 , again gives rise to the minimum / maximum constraints ( [ eq : oom_arg4n3])/([eq : oom_arg4n3a ] ) above .
+ these arguments imply that the rate of change of @xmath307 ( or @xmath308 ) and @xmath309 ( or @xmath310 ) with respect to @xmath12 is sufficiently small at any value of @xmath20 so as to make @xmath12-derivative terms in the euler - lagrange equations negligible - so long as we are restricted to the ranges @xmath311 and @xmath312 , respectively , and so long as the condition @xmath313 holds . in principle , it is possible that the physics of non - cylindrically symmetric strings differs substantially from the symmetric case outside these ranges , but this appears unlikely because the energy density of the field configuration rapidly goes to zero for @xmath314 as @xmath315 .
additionally , we can identify @xmath125 with the correlation length @xmath297 , making @xmath316 a natural cut - off for @xmath20 in accordance with ( [ eq : oom_arg4n3a ] ) .
+ it is also reasonable to assume @xmath317 if we regard the pinch as a kink " or twist " in the string ( analogous with our common sense ideas about ordinary lengths of string ) .
the condition is then equivalent to the statement that it is impossible to twist a length of rope through 360 degrees over a region less than the length of its diameter . in the limit @xmath318 therefore , for all @xmath319 the scalar eom reduces to either @xmath320 or @xmath321 in the limit @xmath322 , . ] for all @xmath323 the vector eom reduces to @xmath324 in the uncoupled regime ( i.e. setting @xmath325 ) , equation ( [ eq : nscs_scalar4 ] ) has the small @xmath20 solution @xmath326 for @xmath319 , which is in accordance with our expectations regarding the effective radius of the scalar core .
likewise equation ( [ eq : nscs_scalar5 ] ) has the small @xmath20 solution @xmath327 which is equivalent .
similarly , in the uncoupled regime ( i.e. setting @xmath328 ) equation ( [ eq : nscs_vector3 ] ) admits the approximate solution for @xmath329 @xmath330 for @xmath331 .
clearly the correct asymptotic solutions in both uncoupled and coupled regimes ( together with the higher order corrections to the small @xmath20 solutions in the latter ) will also be formally analogous to those given for the cylindrical string in the usual literature , under the correspondence ( [ eq : subs1])-([eq : subs3 ] ) .
we need not state them explicitly here . more important for our immediate task
are the results ( [ eq : result1 ] ) and ( [ eq : result2 ] ) , which we now use to to calculate the periodic string tension @xmath18 for the pinched string ansatz . when calculating the approximate @xmath12-dependent tension @xmath18 for the pinched string , we begin by considering the limit @xmath257 , @xmath165 . and @xmath332 so that the field couplings are independent of @xmath32 .
this is equivalent to assuming @xmath333 and @xmath334 for @xmath335 .
] there exist analogues of all the terms which appear in the ( constant ) string tension @xmath19 of a cylindrically symmetric string , according to the identification @xmath336 and @xmath337 where necessary . ]
the @xmath338 and @xmath339 terms of the gradient energy give contributions of the form @xmath340 in this case ( again at leading order ) , exchanging @xmath341 makes no difference to the result of the calculation since all factors of either @xmath53 or @xmath101 exactly cancel in the final step .
in the range @xmath342 the interchange of variables also makes no difference to the form of the contribution from the angular term , which is given by @xmath343 however there now exists a new @xmath344 term in the derivative , which leads to @xmath345 likewise , the contribution from the potential term is formally equivalent under the identification @xmath341 , which gives @xmath346 turning our attention to the gauge field , the contribution from the @xmath12-component of the magnetic flux @xmath347 is analogous to that for the cylindrically symmetric case , so that @xmath348 the only contribution left to calculate comes from the new radial @xmath76-field component , @xmath349 . our new gauge field ansatz is given by @xmath350 so @xmath351 in the limit @xmath335 , we have that @xmath352 for all @xmath12 where @xmath353 and @xmath354 ( i.e. for @xmath355 ) , and @xmath356 for all @xmath12 where @xmath357 ( i.e. at @xmath358 ) .
this implies that the second term inside the brackets is zero for all @xmath12 . and consider the planck - sized regions explicitly . ]
the second term inside the curly brackets vanishes leaving @xmath359 and @xmath360 the first term inside the brackets of ( [ eq : b_r_sqr ] ) yields @xmath361 but this is exactly cancelled by the contribution of the second term @xmath362 so the only remaining contribution is the third term : @xmath363 the final approximate expression for the total string tension is then @xmath364 setting @xmath365 and @xmath366 ( critical coupling ) we then see that the large @xmath32 limit yields @xmath367 we have assumed that the function @xmath368 has period @xmath89 , so that @xmath369 has period @xmath125 .
this implies that @xmath370 , and hence @xmath371 , have period @xmath372 so we may express the latter in the form @xmath373 .
if we also assume that @xmath374 , a simple and natural ansatz for @xmath375 is @xmath376\end{aligned}\ ] ] though , in principle , there are an infinite number of possible ansatz choices . to within an order of magnitude therefore , our final expression for the approximate @xmath12-dependent tension of a pinched string is is that @xmath377 \
\forall z$ ] , we have some freedom to define it to absorb factors of @xmath378 . as mentioned previously ,
the numerical factor in front of the constant term is also to some degree arbitrary and depends upon the definition of @xmath56 .
we therefore assume that both can be adjusted to give @xmath379 where @xmath380 is some function which varies exactly between @xmath381 $ ] . ]
@xmath382.\end{aligned}\ ] ] in the next section we look at @xmath98-strings wrapping cycles around the @xmath383 manifold at the tip of the ks throat . from the lagrangian of the theory
we then determine an approximate formula for the effective four - dimensional tension of the configuration .
we find that , for an appropriate natural ansatz choice for the winding state , this is formally analogous to the result ( [ eq : finally ] ) and this allows us to draw a correspondence between the parameters which define the abelian - higgs model and those which define the tip geometry of ks background .
this ensures that we can connect the string theory model to the field theory one .
for the sake of brevity we refer the interested reader to @xcite for more background on the string theory construction .
we note that cosmic superstrings @xcite have become an important area of cosmological research in recent years .
a selection of papers relating to the observational status of such objects is provided in @xcite-@xcite .
our ansatz ( below ) describes a static string loop of radius @xmath384 with integer windings over all three angular directions of the @xmath383 at the tip of the warped deformed conifold @xmath385 we have chosen to label the loop radius @xmath384 instead of @xmath20 , in order to avoid confusion with the @xmath20-coordinate used in the previous section . + for a string with no intrinsic world - sheet flux @xmath386 ( i.e. @xmath387 , where @xmath388 is the associated momentum ) and adopting canonical coordinates
so that the ansatz above describes non - geodesic windings , the total energy is given by the following potential @xmath389 where @xmath390 denotes the tension of either an @xmath8 or @xmath9-string ( depending on the configuration under consideration ) .
in @xcite the total string mass was separated into a constant piece - corresponding to the mass of the string sections connecting the beads " formed by extra - dimensional windings - and a piece due to the mass of the beads themselves .
it is necessary to set @xmath391 when calculating the mass of an individual bead , but to use @xmath392 when calculating the number of beads in the loop .
this avoids double counting the mass - contribution of the @xmath393-direction windings .
we now perform the same procedure when calculating the effective four - dimensional tension . in order for beads , or fluctuations , in the effective four - dimensional mass - density to occur , we must still set @xmath394 when calculating the approximate mass contained in localised areas ( see @xcite for further details ) . ] + we begin by setting @xmath395 labels the number of windings , in order to avoid confusion with @xmath396 , the topological winding number of the field theory vortex strings . ] in the expression ( [ eq : v1 ] ) above , except in the constant piece ( which is proportional to @xmath397 ) , which again vanishes . taking the limit @xmath398
we then expand our expression for @xmath399 ( to first order ) before dividing through by @xmath400 to obtain @xmath401.\ ] ] we may now make a change of variables via @xmath402 and define a new variable @xmath403 which represents the interbead distance ( or equivalently the interwinding distance , i.e. the four - dimensional length over which a single winding is spread " ) .
this gives @xmath404 which is formally equivalent to ( [ eq : finally ] ) under the identification @xmath405 physically what is happening ( in the string picture ) is the following : when @xmath406 the string is instantaneously wrapping its maximal effective radius @xmath407 in the @xmath383 ( a great circle ) , whereas at @xmath408 the string is instantaneously wrapping the pole of the @xmath383 ( a point ) so that the effective radius of the winding is @xmath409 .
, so that the effective four - dimensional tension is simply @xmath410 . ] at all values of @xmath12 the string wraps some effective radius in the region @xmath411 , and the maximum increase in winding rate with respect to the @xmath12-coordinate , @xmath412 , occurs at the points @xmath413 . from the wound string perspective , these are points at which the greatest length of string is hidden in the compact space ( for a given interval @xmath414 ) , giving rise to maxima in the effective four - dimensional tension .
likewise the periodic part of the tension in the pinched string picture is proportional to @xmath415 and we make the following identification - coordinate of the @xmath383 giving rise to two beads per winding .
if one full winding in the string picture corresponds to one full pinch in the field picture , we would therefore expect to find two peaks in the periodic part of the tension @xmath18 for every one peak in the physical radius of the string core . ]
@xmath416 which explains why we must set @xmath417 in the field theory picture to obtain ( [ eq : correspondence1 ] ) - in this picture there is only one string , which can not give rise to two separate radii .
this allows us to naturally interpret the @xmath12-dependence of the field couplings @xmath10 and @xmath11 via the relation @xmath418 the parameter @xmath419 is fixed and quantised , and @xmath420 represents the fundamental length scale of the theory , so that a string wrapping an effective radius in the @xmath383 , @xmath421 , experiences an effective coupling of approximately @xmath422 this also allows us to establish a relation between @xmath423 , @xmath10 and @xmath11 , which will be investigated shortly .
+ we can imagine a situation where the string completes one full winding in ( say ) the clockwise direction before reversing to wrap the @xmath383 anti - clockwise .
the four - dimensional regions over which these windings take place may correspond to regions of @xmath424 in the field theory picture .
although we still have no effective description for the planck - sized regions of the pinched string in classical field theory , within the string picture we see that there is a relatively natural interpretation , with @xmath425 corresponding to the minimum width of the string due to quantum effects .
+ effectively when the string wraps a point ( at the pole ) of the @xmath383 at @xmath408 , the description of the string as a one - dimensional object breaks down .
it is therefore meaningless to consider the position of the string localised on the @xmath383 on scales smaller than the fundamental string width .
furthermore the four - dimensional effective tension of the string in this region is equal to the intrinsic warped tension @xmath426 , which , under the identification in ( [ eq : correspondence1 ] ) , is equivalent to the tension of a non - cylindrical defect string in the region of the pinch , @xmath427 .
this goes some way towards justifying our original assumption that @xmath428 within the regions @xmath188 $ ] .
+ how tightly wound the string is may affect the four - dimensional length required for it to move from one winding orientation to another . if , for example , we identify the planck length @xmath158 with the fundamental string length @xmath429 , the fact that @xmath430 ( in addition to @xmath431 ) may be explained .
intuitively we expect a more tightly wound string to give a larger value of @xmath32 in the field picture , so that the simple identification @xmath432 must be rejected .
we can , however , arrive at a hypothetical correspondence between @xmath32 and the dynamical parameters which control winding formation in the string picture via the following argument : recall that in the field theory picture we have assumed the sections of neutral " string , which connect neighbouring regions of @xmath424 string , are of length @xmath433
. we may now check this in the dual string picture by assuming that the four - dimensional length over which the string sits " at the pole of the @xmath383 is proportional to ( twice ) the tangent angle of incidence ( i.e. a string which is more tightly wound " requires a greater distance over which to unwind " , in direct proportion to its angle of incidence ) , giving @xmath434 identifying @xmath435 we then have @xmath436 , @xmath437 . using the definition of @xmath438 @xmath439 we see that @xmath440 as @xmath441 .
we therefore propose the identification @xmath442 as @xmath443 and @xmath444 as @xmath440 .
thus , in the limit that we obtain an unwound @xmath8-string , the original duality proposed by nielsen and olesen @xcite is recovered . to be an integer greater than one ( which corresponds to the limit @xmath445 )
we must ensure that @xmath446 , which is equivalent to @xmath447 in the pinched string picture .
hence models in which @xmath448 are potentially problematic , since the assumptions made in order to simplify the pinched string eom break down .
this suggests that further analysis is needed , at least regarding the field - theoretic necklace model . ]
topological winding numbers of opposite signs ( @xmath449 ) way be obtained by taking either physical winding numbers of opposite signs ( @xmath450 ) or opposite signs in front of the square root . + we now consider turning on world - volume flux . for an @xmath8-string ,
this amounts to turning on @xmath9-string charge , and therefore the resulting configuration is a bound - state of @xmath8 and @xmath9-strings , known as a @xmath451 string .
the tension of general @xmath451-string at the tip of this particular warped throat is @xcite-@xcite , @xmath452 so that in the limit @xmath453 , the approximate tension of the @xmath8-string is @xmath454 and that of the @xmath9-string is @xmath455 dealing first with the @xmath8-string , we see that under the identification given in ( [ eq : correspondence1 ] ) ; the only consistent identification between the individual elements @xmath456 and @xmath32 is and @xmath32 directly with string theory parameters , but only through composite expressions of the form @xmath457 ( e.g. @xmath458 as suggested by the expression for the @xmath8-string tension above . ) ] @xmath459 in @xcite it was necessary to identify the energy corresponding to the epoch of the @xmath451-string network formation ( @xmath460 ) with the fundamental string energy scale ( not the warped string energy scale ) so that , @xmath461 not @xmath462 .
although this appears to contradict our identification ( [ eq : f_string_correspondence1 ] ) it is not immediately clear that this is the case since we have chosen to identify the fundamental string width @xmath463 with the fundamental string scale ( so that @xmath464 ) which is in agreement with the previous results .
the identification of @xmath456 with the warped string scale tells us that it is inequivalent to @xmath460 . in the string picture the @xmath98-string network forms at @xmath465 , but the windings only form some time later at @xmath466 where @xmath467 at which point wound strings are dual to defect strings with @xmath468 .
+ the effective formation time may or may not be directly dependent on @xmath469 as this will be determined by the winding formation mechanism and the type of winding formed .
for example , consider the expression for the the total energy of the wound - string loop for geodesic windings formed via the velocity correlations regime , @xmath470 , where @xmath390 denotes the tension of either the @xmath8 or @xmath9-string ( see @xcite ) .
using the arguments given above , geodesic windings in the string picture correspond to un - pinched " ( i.e. cylindrically symmetric ) strings in the dual field theory model , whose total energy we expect to be @xmath471 and whose ( constant ) tension is @xmath472 ( where @xmath473 as in ( [ eq : n_correspondence ] ) ) .
this suggests the correspondence in the wound - string case and @xmath474 in the field theory case . ]
@xmath475 in ( [ eq : f_string_correspondence1 ] ) the symmetry - breaking energy scale is set by the warped string energy scale , whereas in ( [ eq : f_string_correspondence2 ] ) it is set by the fundamental string energy scale .
it is unclear whether the string width should be set by the fundamental string length - scale or the warped string length - scale , and it is similarly unclear whether the energy associated with @xmath8/@xmath9-string network production should be the fundamental string or the warped string energy .
furthermore , it is unclear whether it is possible for the string width to be determined by the fundamental scale ( i.e. @xmath476 ) while the energy associated with network formation ( in the string picture)/symmetry - breaking ( in the field theory picture ) is determined by the warped string scale ( @xmath477 ) , or vice - versa .
clearly therefore , more work is needed to establish the exact relation between the type of wound - string network formed , the winding formation mechanism ( e.g. via a random walk or velocity correlations regime @xcite ) and the effective formation time .
however , we note that the correspondence @xmath478 in both ( [ eq : f_string_correspondence1 ] ) and ( [ eq : f_string_correspondence2 ] ) is encouraging , since in @xcite it was found that , for windings formed via a velocity correlations regime at least , @xmath479 , which is consistent with these results @xcite .
considering the @xmath9-string ( for non - geodesic windings ) also leads to a unique correspondence between the individual field theory and string theory parameters : @xmath480 which preserves this key result .
a relation analogous to ( [ eq : f_string_correspondence2 ] ) also holds for geodesic windings .
+ combining the expression @xmath481 with @xmath482 and ( [ eq : f_string_correspondence1 ] ) gives @xmath483 and hence @xmath484 using the definition of the conifold deformation parameter ( and adjusting the units so that @xmath485 ) , we may then write @xmath486 on the left hand side of ( [ eq : final_correspondence ] ) we have the abelian - higgs couplings , which together with the symmetry - breaking energy scale @xmath456 , determine the effective masses of the particles according to that model ( i.e. the bosons associated with scalar and vector fields ) .
on the right hand side we have string theory parameters which determine the geometry according to the ks model ( including both the large and compact dimensions ) and which control the fundamental mass - scales of the particles associated with that theory ( i.e. the masses associated with the excitations of @xmath8/@xmath9-strings ) .
if we also include @xmath487 , the field theory parameters which set the mass scales for the particles at the current ( post - symmetry breaking ) epoch may be equated with the parameters which control inflation and the cosmological expansion in the string picture .
+ at first glance this may seem strange . however , the decay of the inflaton gives rise to production of particles and defects whose remnants or descendents " ( following temperature changes and further symmetry - breaking phase transitions caused by the expansion and cooling of the universe ) may be observed today . on reflection ,
the identification proposed above appears natural since we expect the parameters which control inflation to set the present day mass - scale of the universe .
+ there are two important remarks to make about the above analysis .
the first regards the nature of the symmetry - breaking phase transition in the string picture . in the abelian - higgs model
the phase transition , giving rise to string formation , is well - defined and it is the @xmath27 symmetry of the vacuum that is broken .
however such symmetry - breaking should be dynamic , and moreover the coupling constants are tuned by hand . according to our argument
, we can equate the parameters determining the size and shape of the mexican hat potential with their string analogues .
but how are we to interpret the symmetry - breaking on the string side ?
we imagine a long , straight ( i.e. unwound ) @xmath98-string obeys @xmath27 symmetry with respect to rotation around its central axis . when windings form this symmetry is broken and each point along the string adopts a phase " factor , determined by it s position in a @xmath488 sub - manifold of the full three - sphere . in the string picture the dynamic nature of the symmetry - breaking process in therefore manifest , and helps us understand why we may write all the parameters which define the field - theoretic strings ( including @xmath32 ) in terms of parameters controlling the @xmath8/@xmath9-string dynamics , and in particular those which control the process of winding formation ( regardless of the exact model of winding formation we adopt ) .
+ the second remark concerns cosmic necklaces , first proposed by matsuda as possible dm candidates @xcite .
the idea was that necklaces which had shrunk to their minimal size ( determined by the fundamental string thickness ) and which contained insufficient mass to undergo collapse , would only interact with other fields / matter gravitationally .
the proposed correspondence between necklace configurations and pinched gauge strings now suggests that this is unlikely to occur , since the presence of a dual @xmath95 term implies that necklaces may also interact with the gauge field . with this in mind
, we propose the following definitions for the effective scalar and vector fields in the wound string model , @xmath489 and @xmath490 where , @xmath200 are subject to boundary conditions analogous to those imposed before . here
@xmath160 may also be interpreted as the angular coordinate of a point on the @xmath491 sub - manifold that defines the effective radius of the winding ( for any value of @xmath12 ) , and @xmath20 is the distance from the centre " of the @xmath491 to its circumference .
physically , a section of wound string effectively occupies a volume @xmath492 in the large dimensions such that @xmath20 and @xmath160 also admit their usual interpretations of the radial and angular coordinates for a string of finite width .
this gives rise to a flux at each point along the string which is quantised in terms of @xmath493 in the string picture ( for fixed values of the other parameters ) and in terms of @xmath396 in the field picture @xmath494 we now use physical arguments to construct dynamic models of pinch formation , which are the direct analogue of dynamical winding formation in the random walk and velocity correlations regimes . to construct the analogue of the former ; consider a field configuration corresponding to an @xmath495-vortex string undergoing random quantum fluctuations in both its phase @xmath160 and magnitude @xmath496 at every point on the string . in order for a section of string within the horizon to undergo a spontaneous transition to a new topological state @xmath497
, the fluctuations in @xmath160 would need to be perfectly correlated over some distance @xmath498 ( i.e. within a volume of approximately @xmath499 ) . for example , a transition from a @xmath118 state to a @xmath119 state , at even a single position @xmath12 corresponds to degenerate vacuum states @xmath500 ( @xmath501 ) tunnelling through the mexican hat potential so that @xmath502 .
this is highly unlikely , regardless of the tunnelling amplitude for such a transition at any individual point .
+ it is also possible for random fluctuations in the magnitude @xmath503 at each point to result in a total reduction of the vortex size to sub - planckian scale .
if this were to occur within the horizon it would be necessary for such fluctuations to be correlated , at least over some length scale @xmath498 , in order for a finite section of string to re - emerge with a different topological winding number . even if this continuously occurs over small regions ( which would be the analogue of continuous brownian motion at every point along a wound string )
, it is clear that the net effect would be to to leave the macroscopic structure of the string , and the total number of pinches , unchanged .
+ the situation is different , however , for the vortex slice which lies ( instantaneously ) on the horizon . in this case
the vortex is free to re - emerge in a differing topological state so that only fluctuations at the horizon may contribute to the net number of pinches .
let us assume that the edge of the vortex at the horizon moves with velocity @xmath504 , but randomly increases or decreases , so that its net velocity may be estimated via @xmath505 where @xmath506 denotes the number of pinches and we have again assumed critical coupling . making the transition to warped space so that @xmath507 and using @xmath508 , the expression for @xmath509 is equivalent to that for @xmath438 : @xmath510 the number of pinches per loop , for a loop formed at @xmath511 is then @xmath512 where @xmath513 is again the step length . setting @xmath514
then results in expressions for @xmath515 , @xmath516 and @xmath517 in the field picture which are the exact analogues of those for @xmath518 , @xmath519 and @xmath520 in the string picture , namely , @xmath521 where we have also identified @xmath522 and @xmath523 and where we have also made the substitution @xmath524 where necessary when transfering between warped and unwarped space .
one may check that under these identifications @xmath525 as @xmath526 , @xmath527 .
though differing from our previous result @xmath528 this is to be expected for windings formed via a random walk where the proportion of the total string length contained in the extra - dimensional windings tends to zero as @xmath526 .
importantly we can now predict the vorticity of the field theoretic strings , in terms of the velocity associated with the fluctuations at the horizon ( or equivalently in terms of the observed number of pinches ) .
if a suitable quantum theory of vortices is developed , it may be possible to predict both @xmath32 and @xmath125 ( or @xmath506 ) in terms of the underlying dynamics .
as a first step towards this goal , note that in the scaling regime , @xmath529 was used to identify @xmath530 , yielding @xmath531 @xcite .
however , we have now constructed a more explicit correspondence between a wound string model in warped space and a defect string in unwarped space which further allows us to identify @xmath532 the physical constraints @xmath533 and @xmath57 , @xmath534 then imply @xmath535 which in conjunction with ( [ eq : izzy ] ) yield @xmath536 , though this is perfectly consistent with the sugra approximation @xmath537 for @xmath538 .
recall that the total mass of a pinched string loop is @xmath539 which , for the ansatz ( [ eq : g_ansatz ] ) , may in turn may be approximated by @xmath540 although further investigation is needed , we content ourselves here with demonstrating an order of magnitude equivalence between the pinched string bead mass and the wound string bead mass . in terms of field - theoretic parameters ,
the bead mass is then @xmath541 where we have used the fact that the mass of an individual bead is half that associated with a single pinch .
we then make use of the recently established identities @xmath542 , @xmath543 , @xmath544 , @xmath545 and @xmath546 to obtain @xmath547 using @xmath548 we then find @xmath549 which is in turn equivalent to , @xmath550 for @xmath528 .
this result is identical to that obtained in @xcite for the mass of a wound - string bead in which the windings were formed via a random walk regime .
however , as mentioned previously , in this case we expect that @xmath551 , rather than @xmath552 ( which we expect to associate with windings formed in a velocity correlations regime ) , which would instead yield @xmath553 there therefore remains a slight inconsistency between the result presented here and those obtained in the previous paper , though it may be hoped that further analysis may be able to clarify this point .
the most important result is that , in either case ( i.e. whether the field - theory expression ( [ eq : bazinga ] ) leads to ( [ eq : omg ] ) or ( [ eq : omg2 ] ) in the string picture ) , the late time fall - off of @xmath554 is proportional to @xmath555 , and the behaviour of pinched string networks may closely mimic that of necklaces formed from wound strings .
+ additionally , we may hope to use the dual pinched string model to answer the question : does bead mass remain fixed after the time of loop formation @xmath556 ? unfortunately , in the absence of a full dynamical model of pinch evolution , we are still unable to answer this question with any certainty .
as the bead mass in this model depends on the quantity @xmath557 we see that two competing factors come into play - the time evolution of @xmath558 in an explicitly dynamic model ) and that of the gradient term @xmath375 . in principle
it is possible for either factor to outweigh the other , so that the bead mass could even increase !
this too may have an analogue in the string picture , whereby the contraction of the string causes neighbouring windings to move closer together , increasing their effective radii .
much more work is needed in both the string and field theory models in order to develop a full dynamical theory .
+ finally , we make a physical observation regarding the construction of a model of pinch formation that is analogous to winding production in the velocity correlations regime . in this case
we assume that the edge of the vortex at the horizon has a classical velocity causing the string core to shrink to the planck scale before growing again to its maximum radius ( reversing direction ) .
dynamically it is unclear why this should be so , but an alternate view of this scenario is that the defect string formed ready - pinched at the epoch of the symmetry - breaking phase transition , so that successive pinches ( separated by a characteristic length scale @xmath125 ) are then simply uncovered by the advancing horizon at a rate proportional to @xmath302 .
we have obtained static vortex solutions to a modified abelian - higgs model which describe non - cylindrical strings by introducing spatially - dependent couplings . hypothesising the existence of planck - sized regions in which vorticity becomes undefined , it was also possible to construct a model in which neighbouring sections of string carry different topological charges . assuming a periodic variation in the pinched string profile , a formal correspondence between the resulting periodic tension and the effective four - dimensional tension of a wound @xmath8/@xmath9-string in the ks geometry was obtained . using a specific , but natural , ansatz choice for the string embedding describing non - geodesic windings
, we were able to obtain specific identities between string theory parameters ( defining the ks geometry ) and the field theory parameters ( defining the abelian - higgs model ) . in the dual string picture ,
the spatial dependence of the field couplings were found to be related to the effective radius of the windings .
+ one interesting observation about this matching is that the field theory result is at large @xmath32 ( i.e. large winding number ) , whilst the @xmath451-string tension has large flux @xcite .
more concretely the matching on the field theory side was through the parameter pairing @xmath559 , with @xmath560 on the string theory side , which suggests a relation between the symmetry breaking scale ( set by @xmath456 ) and the string length .
+ though there are many possible ways in which to improve and build upon this work , the most valuable would be to extend the present analysis from the purely static case to the more general dynamical one .
ideally complete dynamical models of pinched string formation ( and pinch evolution ) would be developed which would allow us to determine the cosmological consequences of pinched string networks with greater accuracy .
it would also be instructive to compare these with general dynamical models of wound @xmath451-strings in order to determine if correspondences exist for more general string backgrounds .
a simple example would be the inclusion of fluxes present in the warped deformed conifold solution , but analysis in more general backgrounds would be most likely to lead to a better understanding of the field - string correspondence .
moreover we have only considered models of string necklaces in this paper , and the extension to more general string lattice configurations ( where each bead is attached to @xmath561 strings ) remains to be developed @xcite there is significant scope for future work , including the matching of such models to the cmb anisotropy data and evaluating the projected gravitational wave spectrum .
we intend to return to such topics in future .
the interested reader is referred to @xcite for a more detailed discussion of these topics .
n. bevis , m. hindmarsh , m. kunz and j. urrestilla , _ cmb power spectrum contribution from cosmic strings using field - evolution simulations of the abelian higgs model _ , _ phys .
* d 75 * ( 2007 ) 065015 [ astro - ph/0605018 ] [ spires ] .
w. h. zurek , _ cosmological experiments in superfluids and superconductors _ , prepared for _ nato advanced study institute on formation and interactions of topological defects _ , cambridge , england , 21 aug-3 sep 1994 [ spires ] .
m. j. bowick , l. chandar , e. a. schiff and a. m. srivastava , _ the cosmological kibble mechanism in the laboratory : string formation in liquid crystals _ , _ science _ * 263 * ( 1994 ) 943 [ hep - ph/9208233 ] [ spires ] . p.c .
hendry , n.s .
lawson , r.a.m .
lee , p.v.e . mcclintock and c.d.h .
williams , _ creation of quantized vortices at the lambda transition in liquid helium-4 _ , _ j. low temp .
_ * 93 * ( 1995 ) 1059 [ cond - mat/9502119 ] [ spires ] .
|
we modify the standard abelian - higgs model by introducing spatially - dependent couplings for the scalar and vector fields .
we investigate static , non - cylindrically symmetric solutions of the resulting field equations and propose a pinch solution which interpolates between degenerate vacua along the string , labelled by @xmath0 .
this configuration corresponds to a vortex which shrinks to the planck scale before re - emerging as an anti - vortex , resulting in the formation of a bead pair with one bead either side of the intersection .
the solution is then topologically stable .
a key assumption is that quantities such as phase and winding number , along with those which depend on them like the magnetic flux , become undefined at the planck scale so that regions of opposite winding may be joined via a planck - sized segment of neutral string .
similarities between this solution and the extra - dimensional windings of strings in type iib string theory are discussed and a correspondence between field theory and string theory parameters is suggested . the spatial - dependence of the field couplings is found to have a natural interpretation in the string picture and results from the variation of the winding radius , giving rise to a varying ( effective ) string coupling .
an interesting result is an estimate of the higgs mass ( at critical coupling ) in terms of the parameters which define the klebanov - strassler geometry and which , in principle , may be constrained by cosmological observations . @xmath1 @xmath2 _ @xmath3 center for research in string theory , queen mary , university of london + mile end road , london e1 4ns , uk + _ _ @xmath4 astronomy unit , school of mathematical sciences , queen mary , university of london + mile end road , london e1 4ns , uk + _ _ @xmath5 department of physics and astronomy , university of victoria , victoria , bc + v8p 1a1 , canada _
|
in the last decade much attention has been focused on the possibility that a sizable fraction of galactic dark matter consist of machos ( massive astrophysical compact halo object ) . since 1992 , the macho ( alcock et al . 1993 ) and
eros ( aubourg et al . 1993 ) collaborations have looked towards the large and small magellanic clouds ( lmc and smc ) in order to detect machos using gravitational microlensing .
this technique , originally proposed by paczyski ( 1986 ) , analyses the luminosity variation of resolved source stars , due to the passage of machos close to the line of sight between the source and the observer . the macho collaboration ( alcock et al . 2000 ) discovered 13 - 17 microlensing events towards the lmc .
assuming that all events are due to machos in the halo , about @xmath0 of the halo dark matter resides in form of compact objects with a mass in the range @xmath1 @xmath2 .
the eros collaboration ( lasserre et al . 2000 ) observed 6 microlensing events , 5 in the direction of the lmc and 1 in the direction of the smc .
these observations place an upper limit on the halo dark matter fraction in the form of machos . in particular ,
they exclude , at the @xmath3 confidence level , that more than @xmath4 of a standard halo is composed of objects in the range @xmath5 @xmath6 .
note that the results of the two collaborations are consistent with a @xmath7 halo dark matter fraction of objects @xmath8 .
the ogle collaboration ( udalski et al . 1993 ) originally searched for microlensing events only towards the galactic bulge , but has now also extended its search to the lmc and smc .
a natural extension of the microlensing observational technique consists in observing dense stellar fields even if single stars can not be resolved , as in the case of the m31 galaxy . for this purpose ,
the pixel lensing technique has been proposed ( baillon et al . 1993 ) and then implemented by the agape collaboration ( ansari et al .
another technique , based on image subtraction , has been developed by the vatt - columbia collaboration ( crotts 1992 ; tomaney & crotts 1994 ) , and is used also in the wecapp project ( riffeser et al .
the monitoring of m31 has the advantage that the galactic halo can be probed along a line of sight different from those towards the lmc and smc .
furthermore , the observation of an external galaxy allows one to study its halo globally , which , in the case of m31 , has a particular signature due to the tilted disk .
accordingly , the expected optical depth for microlensing varies from the near to the far side of the m31 disk ( crotts 1992 ; jetzer 1994 ) .
the efficiency of the pixel lensing method to detect luminosity variations has been tested by the agape collaboration on data taken at the 2 meter bernard lyot telescope in two bandpasses ( @xmath9 and @xmath10 ) , covering 6 fields of @xmath11 each around the center of m31 , during 3 years of observations ( 1994 - 1996 ) .
a possible microlensing candidate has been observed and further characterized by using information from an archival hubble space telescope wfpc2 image ( ansari et al .
an important conclusion of this analysis is that it is crucial to collect data in two bandpasses over a long duration with regular sampling .
very recently , the point agape collaboration ( aurire et al . 2001 ) announced the discovery of a short timescale candidate event towards m31 .
additional microlensing candidates towards the same target have been reported by the vatt - columbia collaboration ( crotts et al .
2000 ) . in this paper
, we present results for the 1998 - 1999 campaign of observations at the 1.3 meter mcgraw - hill telescope , mdm observatory , kitt peak , towards the andromeda galaxy . in @xmath12 [ pixellensing ]
, we briefly outline the pixel lensing technique . @xmath12
[ setup ] is devoted to the description of the observational campaign and the experimental setup . in @xmath12 [ reduction
] we discuss the data reduction procedure ( calchi novati 2000 ) in some detail , in particular the approach used to eliminate instabilities caused by the seeing and to evaluate the errors . in @xmath12 [ anaml ]
we present our selection pipeline ( calchi novati 2000 ) : bump detection ( @xmath12 [ anabump ] ) , shape analysis ( @xmath12 [ anashape ] ) and color and timescale selection ( @xmath12 [ anamira ] ) .
we select a sample of 5 light curves that we retain as microlensing candidate events and whose characteristics are given in @xmath12 [ anaevts ] . in @xmath12 [ novae ]
we show the light curve of a nova located inside our field of observation : the discovery of variable sources is a natural byproduct of the microlensing search . in @xmath12 [ mcs ]
we conclude with a comparison of the outcome of our selection with the prediction of a monte carlo simulation .
pixel lensing is an efficient tool for searching microlensing events when the sources can not be resolved . in this case , the light collected by each pixel is emitted by a huge number of stars .
although , in principle , _ all _ stars in the pixel field are possible sources , one can only detect lensing events due either to bright enough stars or to high amplification events .
typical sources are red giant stars with @xmath13 $ ] .
we estimate that there are about @xmath14 sources per square arc second that fit these requirements .
the main drawback of the method is that usually we have no direct knowledge of the flux of the unamplified source . for images obtained by the observation of dense stellar fields ,
the flux collected by a pixel is the sum of the fluxes emitted by single stars , which all contribute to the background .
if one of these stars is lensed , its flux varies accordingly . whenever this variation is large enough , it will be distinguishable from the background produced by the other stars . denoting by @xmath15 the amplified flux detected by the pixel and by @xmath16 the background flux , the flux variation is given by @xmath17\,,\ ] ] where @xmath18 represents the amplification as a function of time , @xmath19 @xmath20 is the flux of the star before lensing , and @xmath21 is that given by the other stars . in the point - like and uniform
motion approximations , @xmath22 is related to the lensing parameters by : @xmath23 @xmath24 where @xmath25 is the _ einstein time _ , @xmath26 the impact parameter in units of the _ einstein radius _ @xmath27 , and @xmath28 the time of the maximum amplification .
the einstein radius is @xmath29 where @xmath30 is the mass of the lens , @xmath31 , @xmath32 and @xmath33 are the observer - lens , observer - source , and lens - source distances , respectively .
two characteristic features of a microlensing event are achromaticity and the uniqueness of its luminosity bump , although differential amplification of extended sources can give rise to a chromatic , but still symmetric , lensing light curve ( han et al .
the data analysed in this paper have been collected on the 1.3 meter mcgraw - hill telescope , at the mdm observatory , kitt peak ( usa ) .
two fields have been observed , which lay on the two sides of the galactic bulge ( see fig .
[ fig : plotm31 ] ) and have been chosen in order to be able to study the expected gradient in the optical depth .
the two fields are almost parallel to the major axis of m31 ; their centers are located at @xmath34 00h43m24s , @xmath35 ( j2000 ) ( named `` target '' ) , and @xmath34 00h42m14s , @xmath36 ( j2000 ) ( named `` control '' ) .
the data acquired in the target field are analysed here .
[ fig : plotm31 ] shows the location of the fields and for comparison also the smaller agape field .
the observations were taken with a ccd camera of @xmath37 pixels with @xmath38 and therefore a total field size of @xmath39 . in order to test for achromaticity
, images have been taken in two bands , a wide @xmath10 and a near standard @xmath40 .
the exposure time is 6 minutes for @xmath10 , 5 minutes for @xmath40 .
the observations started in the fall of 1998 and are still underway .
here we analyse the data taken in the period from the beginning of october 1998 to the end of december 1999 . in fig .
[ fig : images ] we give the time sampling of the measurements ( number of nights and images ) . for each night
we have measurements in both @xmath10 and @xmath40 . on average , there are twice as many @xmath10 images as @xmath40 images . in @xmath10 band
we have @xmath41 images distributed along 42 nights of observation .
most of the observations ( about @xmath42 ) are concentrated in the first three months , so that , unfortunately , the time distribution of the data is not optimized for the study of microlensing effects .
thus , the given time distribution allows us to select events that take place almost exclusively during the first three months of observation .
furthermore , the time coverage of about 14 months is still not long enough to test conclusively the bump uniqueness requirement for a microlensing event .
mainly for this reason , we will speak in this paper only of _ candidate _ microlensing events . taking into account the transmission efficiency of the filters and the catalogued magnitudes @xmath43 and @xmath44 ( cousins colour system ) , for a sample of @xmath45 reference secondaries identified in the target field ( magnier et al .
1993 ) , we derive the following photometric calibrations @xmath46 @xmath47 where @xmath48 and @xmath49 is the flux of the source in @xmath50 measured in the @xmath10 and @xmath40 filters respectively .
the estimated error is @xmath51 mag , both for @xmath10 and @xmath40 .
during each night of observation about 20 images are taken in @xmath10 and 12 in @xmath40 . in principle , this allows two possible strategies for the analysis .
we can study flux variations on light curves built either by a point obtained from each image , or by a point obtained by averaging over many images . in the first case ,
we are potentially sensitive to very short time variations . however , this sensitivity is undermined by the low signal - to - noise ratio ( s / n ) . in the second case , the s
/ n is increased by the square root of the number of images we combine .
results of the analysis of light curves built with one point per image will be discussed in a future paper . here
we concentrate on the analysis of light curves obtained after combining all the images taken in the same night , using a simple averaging procedure performed on geometrically aligned images ( see below ) .
data reduction is carried out as follows .
after the usual corrections for instrumental effects , debiasing and flatfielding , we normalize all the images to a common reference to cope with variations induced by the observational conditions which are different from image to image ( so that we get _ global _ stability conditions on each image with respect to a given one ) .
we can distinguish three separate effects : the geometric offset of each image with respect to the others , the difference in photometric conditions of the sky and seeing effects . by means of geometrical alignment
we obtain that each pixel , on all the images , is directed towards the same portion of m31 .
we take advantage of the fact that the mean seeing disk is much larger than the pixel size .
we follow ansari et al .
( 1997 ) , and get a precision better than @xmath52 . following the methods developed by the agape collaboration ( ansari et al .
1997 ) , we then bring all the images to the same photometric conditions in such a way that the images are globally normalized to a common reference .
the procedure is based on the hypothesis that a linear relation exists between `` true '' and measured flux .
it then follows that @xmath53 where @xmath54 and @xmath55 are the fluxes in pixel @xmath56 for the reference and the current image , and @xmath57 and @xmath58 are two correction coefficients , which are the same for all the pixels of the image , that take into account the effects of variable atmospheric absorption and variable sky respectively .
the seeing effect gives rise to a spread of the received signal .
in our data the seeing varies from @xmath59 up to @xmath60 .
consequently we observe fake fluctuations on light curves obtained after photometrical alignment . in order to cope with this effect , and thus to get reasonably stable light curves ,
we follow a two steps procedure ( for further details see ansari et al .
( 1997 ) and le du ( 2000 ) ) .
we begin by substituting the flux of each pixel with the flux of the corresponding _ superpixel _ , defined as the flux received on a square of @xmath61 pixels around the central one .
the value @xmath62 should be large enough to cover the typical seeing disk , but not too large to avoid an excessive dilution of the signal . given the mean seeing value and the angular size of the pixel , we choose @xmath63 .
this corresponds to @xmath64 , compared to the average value for the seeing of @xmath65 for both @xmath10 and @xmath40 images . in this way
we get a substantial gain in stability since elementary pixels are strongly affected by seeing fluctuations . denoting by @xmath66 the flux in a superpixel , we have @xmath67 where @xmath68 and @xmath62 is the superpixel size . instead of trying to evaluate the point spread function of the image , as a second step we apply an empirical stabilization of the difference between the flux measured on the image and that of the _ median _ image , obtained by removing small scale variations with a median filter on a very large window of @xmath69 pixels ( in this way we get an image whose signal is independent from the seeing value ) .
the stabilization is then based on the observed linear correlation , for each superpixel , between these differences measured on the current image ( after photometrical alignment ) and the reference image . denoting with @xmath70 and @xmath71 the value of the flux in a superpixel @xmath56 for the reference , the current ( photometrically aligned ) and the median images respectively
, we have the empirical relation @xmath72\,.\end{aligned}\ ] ] the slope @xmath73 , calculated with a minimization procedure , shows a clear correlation with the seeing ( fig . [
fig : alpha ] ) .
this variation is expected because the flux of a star that enters a superpixel changes with the seeing . in the example shown in fig .
[ fig : cigar ] the seeing of the current image is greater than that of the reference image , and we find @xmath74 .
we note that in this case , the flux of superpixels for which @xmath75 are corrected to a lower value ( points in the bottom left corner ) , while , if @xmath76 , they are corrected up ( points in the up right corner ) .
if the seeing of the current image is smaller than that of the reference image ( @xmath77 ) the situation is reversed .
while stressing its empirical character , we note that this approach is rapid and efficient . in fig .
[ fig : seeingcl ] we show the effect of the correction on a given light curve .
to construct a corrected current pixel flux as close as possible to the reference flux , @xmath78 , we replace @xmath78 in ( [ seeingcorrection ] ) by the corrected current flux @xmath79 and solve for the latter : @xmath80\,.\end{aligned}\ ] ] that is , the corrected flux for the current image is given by the sum of the median of the reference image , and of the weighted deviation from its median , i. e. by its characteristic small spatial scale variations , depending on the relative absorption and on the seeing conditions ) with @xmath81 replaced by @xmath82 ( i.e. using non photometrically aligned images ) in order to perform in one single step the photometrical alignment and the seeing correction .
in the two cases we get the same final results . ] . in order to minimize the deviations from the median , we choose as a reference image ( one for each filter ) , an image characterized by a seeing value equal to the average value over the period of observations . the seeing fraction of a source in a superpixel is then @xmath83 .
another crucial point is the evaluation of the error to be associated with the received flux . in order to give a more appropriate , though empirical , error estimate , we renormalize the photon noise @xmath84 by introducing a correction factor @xmath85 , that depends on the image , to include all the systematic effects over which we have less control .
for a discussion on the relation between the photon noise and others different systematic effects such as surface brightness fluctuations see gould ( 1996 ) . the evaluation of @xmath85 is based on the study of the dispersion of the distribution of the normalized difference , superpixel by superpixel , between the current and the reference image @xmath86}\,,\ ] ] where the denominator is the statistical error on the difference .
this distribution is expected to have zero mean ( which follows from the geometrical and photometrical alignment ) and dispersion one ( which would indicate that the photon noise alone gives the right evaluation of the error ) .
we do find a null mean but the dispersion is greater than one and depends on the seeing value .
we note , however , that this effect is greatly reduced by the seeing correction ( see fig .
[ fig : dispe ] ) . for the reference image
we assume that the estimated error @xmath87 $ ] coincides with the statistical error given by the photon noise , @xmath88= \sigma^2_{stat}\left[\phi_{ref}(i , j)\right]\,,\ ] ] because it does not suffer from additional noise due to geometric , photometric and seeing transformations , as do the other images . on the other hand
, this means that we evaluate the estimated error with respect to that of the reference image .
the correction factor @xmath85 is then equal to the dispersion of the distribution ( [ gaussiana ] ) calculated for a sample of points , properly selected according to the criterion that they belong to `` stable '' light curves , in order to exclude light curves which show real stellar flux variations .
if the subset is small enough and homogeneous , the correction factor appears to be , as expected , almost independent of the seeing value , and has an average value @xmath89 ( fig .
[ fig : dispe ] , bottom ) .
the estimated error of the @xmath56 superpixel is obtained from the relation @xmath90= \nonumber \\ \eta^2_{curr}\,\cdot\sigma^2_{stat}\left[\hat{\phi}_{curr}(i , j)-\phi_{ref}(i , j)\right]\,.\end{aligned}\ ] ] the `` stable '' pixels are selected by imposing the condition @xmath91 } \le \epsilon\,,\ ] ] where @xmath92 is the average flux along the light curve , @xmath93 is the _ baseline _ level and @xmath94 $ ] is the error associated with the evaluation of the averages .
the baseline is defined as the minimum value taken by the average flux calculated along @xmath95 consecutive points on the light curve .
we take @xmath96 , in order to avoid underestimation of the baseline due to fluctuations .
choosing @xmath97 , the selected points are @xmath98 of the total .
a microlensing event is characterized by specific features that distinguish it from other , much more common , types of luminosity variability , the main background to our search . in particular for a microlensing event the bump * does not repeat ;
* follows the ( symmetric ) paczyski shape ; * is achromatic .
variable stars usually show multiple flux variations and have an asymmetric chromatic shape .
moreover , different classes of variable stars are characterized by specific features , such as timescale variation and color , that can be used to distinguish them from real microlensing events . in the following we devise selection techniques that make use of these characteristics while taking account of the specific features of our data set . as a first step we select light curves showing a single flux variation .
we begin by evaluating a baseline , i.e. the background flux ( @xmath99 ) along each light curve , as defined in @xmath12 [ reduction ] .
once the baseline level has been fixed , we look for a significant bump on the light curve .
this is identified whenever at least 3 consecutive points exceed the baseline by @xmath100 .
the variation is considered to be over when 2 consecutive points fall below the @xmath100 level . under the hypothesis
that the points follow a gaussian distribution around the baseline , we use the estimator @xmath101 , the likelihood function , to measure the statistical significance of a bump .
we want to give more weight to points that are unlikely to be found , so that we define @xmath101 as for the corrected superpixel flux @xmath79 as given in equation [ seeingcorrected ] . ]
@xmath102 where @xmath103.\ ] ] @xmath101 is then a growing function of the unlikelihood that a given variation is the product of random noise .
this estimator is different from the usual definition leading to a @xmath104 with @xmath105 points , and has the advantage to give weight only to the positive deviations above threshold , which are the ones of interest . for each light curve
we denote by @xmath106 and @xmath107 the two largest deviations , respectively .
we fix a threshold @xmath108 , and we require @xmath109 to distinguish real variations from noise .
moreover , we fix an upper limit to the ratio @xmath110 to exclude light curves with more than one significant variation .
the shape analysis is then carried out on the superpixels that have the highest values of @xmath101 in their immediate neighborhood since we find a cluster of pixels associated with each physical variation .
this method suffers from a possible bias introduced by an underestimation of the baseline level ( which we further analyse in the next section ) .
we have carried out a complete analysis selecting the pixels with the following criteria : * exclusion of resolved stars ; * @xmath111 ; * @xmath112 .
this selection is made only on @xmath10 images in order to reduce contamination by variable sources . in this way
we take advantage of the fact that most luminous variables ( to which we are anyway sensitive ) show stronger variations in the @xmath40 than @xmath10 band . by using these peak detection criteria ,
the number of superpixels is reduced from @xmath113 to @xmath114 . as a second step we determine
whether the selected flux variation is compatible with a microlensing event .
the light curve of a microlensing event with amplification @xmath18 due to a source star with unlensed flux @xmath20 ( now to be evaluated in a superpixel ) is @xmath115 where @xmath116 represents the flux collected in the superpixel associated with a single pixel , as defined before , and @xmath18 is given by ( [ amplificazione ] ) .
actually , one can not directly and easily measure @xmath20 , the unamplified flux of the unresolved source star .
only a combination of the 5 parameters that characterize the light curve can be measured in a straightforward manner : * @xmath117 , the background level ( which include the flux of the unamplified source ) ; * @xmath118 , the time of maximum amplification ; * @xmath119 , the time width of the bump at half
maximum ; * @xmath120 , the excess of the flux with respect to the background at maximum . whenever the amplification is high enough , one can approximate @xmath121 and @xmath122 .
it is then possible to rewrite the expression ( [ lightcurve ] ) in terms of these 4 parameters , and a degeneracy arises among the parameters of the amplification @xmath123 and @xmath124 , and the unknown flux of the unamplified source , @xmath125 ( `` degenerate '' paczyski curve gould ( 1996 ) ) .
because of this degeneracy it is in general difficult to get , without extra - information , a reliable insight into all the parameters that characterize the light curve , the einstein time in particular .
for this reason we can extract only the 4 aforementioned parameters even though we carry out a non linear fit with the complete 5 parameters ( `` non - degenerate '' ) paczyski curve .
we now refine the selection based on the likelihood estimator in order to remove unwanted light curves with low s / n and for which the available data do not allow us to well characterize the bump . to this end we perform a paczyski 5parameters fit and we study * the signal to noise ratio for the @xmath10 flux variation ; * the sampling of the data on the bump .
we define the s / n estimator as @xmath126 where @xmath127 is the @xmath128 with respect to a constant flux and @xmath129 is the @xmath128 with respect to the paczyski fit . the ratio @xmath130 is actually correlated with the likelihood estimator @xmath101 we used in the previous step . in parallel with the cut
@xmath131 we then keep only light curves with @xmath132 .
we do not ask for the @xmath40 bump to be significant .
the second point concerns the necessity to well characterize the bump shape in order to recognize it as a microlensing event in the presence of highly irregular time sampling of data ( see fig .
[ fig : images ] ) . for this purpose
we require at least 4 points on both sides of the maximum , and at least 2 points inside the interval @xmath133 .
after this selection we are left with 1356 flux variations . from now on we work with the data in both colors ( @xmath10 and @xmath40 )
and we carry out a shape analysis of the light curve based on a two steps procedure as follows : * @xmath128 selection criterion ; * durbin - watson test on residuals ; which we now discuss in some detail .
the first point is taken into account by performing the non - degenerate paczyski fit in both colors simultaneously , so that we check also for achromaticity of the selected luminosity variations .
in particular , we require that the three geometrical parameters that characterize the amplification ( @xmath134 and @xmath28 ) be the same in both colors . we get , therefore , a 7 parameters least @xmath128 non linear fit : @xmath135 ^ 2}{\sigma_{i , j}^2}\,.\ ] ] to retain a light curve as a candidate microlensing event , we require that the reduced @xmath128 @xmath136 where @xmath137 is the total number of points in @xmath40 and @xmath10 .
the application of the @xmath128 criterion test reduces the sample of light curves from 1356 to 27 . as a further step
we apply the durbin watson ( durbin & watson 1951 ) test to the residuals with respect to the 7-parameters non - degenerate paczyski fit . with the dw test we check the null hypothesis that the residuals are not timely correlated by studying possible correlation effects between each residual and the next one against type i error ( i.e. against the error to reject the null hypothesis although it is correct , e.g. babu & feigelson ( 1996 ) ) .
we require a significance level of 10% . as _ time _ plays a fundamental role in the dw test , we perform this test on each color separately .
we call @xmath138 and @xmath139 the coefficients for the durbin watson test on the full data set . in order to retain a light curve we require @xmath140 , appropriate for 4042 points along the light curve .
this statistical analysis reduces our sample of light curves from 27 to 11 .
it is worthwhile to note that some light curves , showing a real microlensing event superimposed on a signal due to some nearby variable source , and passing the previous selection criteria , could be excluded by the dw test , sensitive to timely correlated residuals . in order to test our efficiency with respect to the introduction of the dw test we have done a study on `` flat '' light curves performing a `` constant flux '' fit , and selecting light curves requiring a reduced @xmath141 . by applying the dw test
we then reject about 50% of these light curves , i.e. , more than the 10% we could expect if we had just random fluctuations . in the discussion of the monte carlo simulation
we duly take into account this effect . [
cols="^,^,^",options="header " , ] by far , the most efficient way to get rid of multiple flux variations due to variable stars is to acquire data that are distributed regularly for a sufficiently long period of time .
unfortunately , at present , the data cover with regularity only the first three months of observation , and the total baseline is less than 2 years long .
for this reason , in addition to the analytical treatment that looks for the compatibility with an achromatic paczyski light curve ( efficient , for instance , to reject nova - like events ) , we introduce another criterion based on the study of some physical characteristics of the selected flux variations .
in particular we note that long period red variable stars ( such as miras ) could not be completely excluded by the selection procedure applied so far . by contrast
, short period variable stars are eliminated thanks to the cut on the second bump of the likelihood function . a preliminary analysis of the period , the color and the light curves of long period red variable stars , taken from de laverny et al .
( 1997 ) , lead us , with a rather conservative approach , to exclude those light curves that present at the same time a duration @xmath142 days and a color @xmath143 . a more detailed analysis aimed at a better estimation of that background noise
is currently underway .
we are aware that this last selection criterion could eliminate some real microlensing events .
the probability that this might happen is however low because the microlensing timescales are expected to be uncorrelated with the source color .
in fact , a combination of a large @xmath144 and color @xmath145 is extremely unlikely for microlensing events , but quite common for red variables .
this last criterion further reduces the number of candidates from 11 down to 5 .
we now summarize ( table [ tavola1 ] ) the different steps in the selection , give the number of surviving pixels after the application of the indicated criteria and the details of our set of microlensing candidate events .
we take these 5 light curves , whose characteristics we are now going to discuss , as our final selection of microlensing candidate events . in the following table ( table [ tavola2 ] )
we give their position , the estimated @xmath144 in days , the time of maximum amplification @xmath118 ( j-2449624.5 ) , the magnitude at maximum . ] @xmath146 and the color @xmath147 .
we then give the values of the fit : the reduced @xmath148 and the values of the durbin
watson @xmath149 coefficients .
the corresponding light curves are shown in fig .
[ fig:5evts ] .
our data contain many more varying light curves that are due not to microlensing events but to other variable sources .
the study of these variable stars is an interesting task in itself .
clearly , pixel lensing is well suited for this research .
we give here ( see fig . [
fig : novacl ] ) only the light curve of an event characterized by a very strong and chromatic flux variation , that has already been considered as due to a nova ( modjaz & li 1999 ) and whose light curve is also shown in riffeser et al .
we note that our points , even if with a poorer sampling in the descent , are in good agreement with those of this collaboration .
this nova is found in the region of the bulge of m31 and we localize it in @xmath150=00h43m1.7s,@xmath151 ( j2000 ) .
we evaluate the magnitude at maximum amplification as @xmath152 and @xmath153 .
in order to gain an insight into the results obtained , we compare our 5 candidate microlensing events with the prediction of a monte carlo simulation which takes into account the experimental set up and the time sampling of the observations .
we assume a standard model ( isothermal sphere with a core radius of 5 kpc ) for the haloes of both m31 and the milky way .
the total mass of m31 is assumed to be twice that of our galaxy .
machos can be located in either haloes . moreover , we consider also self lensing due to stars in the m31 bulge or disk .
we fix the lens masses at different values for machos in the halo and stars in the bulge or disk .
the model of the bulge is taken in kent ( 1989 ) , the luminosity function in han et al .
the luminosity function of the disk is determined considering two models : the one developed in devriendt et al .
( 1999 ) and the model obtained considering the data of the solar neighborhood taken in allen ( 1973 ) corrected for high luminosities ( hodge et al 1988 ) .
the results we obtain are almost insensitive to the particular choice between these two models .
we choose the mass of the lenses in the bulge to be @xmath154 , and the mass for machos in the haloes equal either to @xmath155 or to @xmath156 . in both cases about @xmath157 of the expected events
are due to lenses located in m31 .
taking into account our selection criteria , the results for the expected number of events for a halo fully composed of machos , including a contribution due to lensing by stars of the bulge and disk of m31 of @xmath158 event independent of halo parameters , is @xmath159 and @xmath160 , respectively .
the monte carlo simulation does not yet include the effect of secondary bumps due to artifacts of the image processing ( alignment , seeing stabilization ) and to underlying variable objects . from the data
, we estimate that these effects reduce the number of observed events by at most 30% , and this for the shortest events .
we expect , and this is confirmed by simulations , that the sources of most detectable events are red giants and have very large radii .
therefore , finite size and limb darkening effects are important , in particular for low mass lenses .
these effects are included in the simulations .
however , in both the real and simulated analysis , we do not include finite size and limb darkening effects in the amplification light curve fitted to the events .
this results in a loss of detection efficiency smaller than 5% , which is taken into account in the simulations .
locating our candidates in the parameter space predicted by the simulation is more meaningful .
we report the expected values of @xmath144 and on the @xmath10 magnitude at maximum . in figs .
[ fig : mc001 ] and [ fig : mc05 ] we give plots of their functional relation and of their projected distributions . on these same plots
we give the position of our selected light curves in this parameter space .
looking at the distributions we notice that for the @xmath155 and @xmath156 cases , @xmath42 of the light curves are expected to have a time width at half maximum @xmath161 and @xmath162 days , respectively ( of course , shorter events are expected when the macho mass is smaller ) . in both cases , @xmath163 of the events are predicted with a magnitude at maximum @xmath164 .
our candidates have @xmath165 days and @xmath166 ( somewhat at the limit of the expected distributions ) and therefore most of them are probably not microlensing events .
still , we expect @xmath158 self lensing event and it is possible that one or two of them are true microlensing events . in any case , from the @xmath144 distribution , we are led to exclude that the microlensing candidate events are due to machos of very small mass ( only @xmath167 of events with @xmath156 are expected to have @xmath168 days ) .
this is indeed in agreement with the results found by the macho and eros collaborations : they find lens masses in the halo within the range 0.20.6 @xmath169 ( alcock et al .
2000 ; lasserre et al . 2000 ) .
we are not yet in a position to tell what kind of varying objects generate our events if they are not due to microlensing
. they may be irregular or long period variable giants , but only a longer time baseline , and/or observations of the object at minimum light , will allow us to conclude . the mdm analysis is not yet complete . the results from the analysis of data acquired in the other field ( located on the opposite side with respect to the major axis of m31 of the target field analysed here ) and results from new observations scheduled for
the fall 2001 will give us the opportunity to gain further insight into the still open question of the composition of dark haloes .
at the present time , with the caution suggested by the just mentioned problems , the analysis discussed in this paper tends to confirm that only a minor fraction of dark matter in the galactic haloes is in form of machos within the mass range 0.010.5 @xmath169 under the assumption of a standard model of the halo and given source luminosity functions .
we thank m. crz , s. droz , l. grenacher , g. marmo , g. papini and n. straumann for useful discussions and suggestions .
work by ag was supported by nsf grant ast @xmath17097 - 27520 and by a grant from le centre franais pour laccueil et les echanges internationaux . 99 alcock , c. , et al .
1993 , nature , 365 , 621 alcock , c. , et al .
2000 , apj , 542 , 281 allen , c. w. , ed . 1973 ,
astophysical quantities ( london , athlone press ) alonso , a. , et al . 1999 , a&as , 140 , 261 ansari , r. , et al .
1997 , a&a , 324 , 843 ansari , r. , et al .
1999 , a&a , 344 , l49 .
aubourg , e. , et al .
1993 , nature , 365 , 623 aurire , m. , et al .
2001 , apj , 553 , l137 babu , g. j. , & feigelson , e. d. 1996 , astrostatistics ( chapman & hall , london ) baillon , p. , et al .
1993 , a&a , 277 , 1 calchi novati , s. 2000 , microlensing gravitazionale per la rivelazione di machos in direzione della galassia m31 : analisi dati con il metodo di agape , phd thesis , university of salerno crotts , a. p. 1992
, apj , 399 , l43 crotts , a. p. , et al .
2000 , astro - ph/0006282 de laverny , p. , et al .
1997 , a&as , 122 , 415 devriendt , j. , guiderdoni , b. , & sadat , r. 1999 , a&a , 350 , 381 durbin , j. , & watson , g. s. 1951 , biometrika , 38 , 159 gould , a. 1994 , apj , 421 , l71 gould , a. 1996 , apj , 470 , 201 han , c. , jeong , y. , & kim , h. 1998 , apj , 507 , 102 han , c. , park , s. , & jeong j. 2000 , mnras , 320 , 41 hodge , p. , lee , m. g. , & mateo , m. 1988 , apj , 324 , 172 jetzer , ph .
1994 , a&a , 286 , 426 kent , s. m. 1989 , aj , 97 , 1614 le du , y. 2000 , agape : leffet de microlentille gravitationnelle pour la recherche de matire noire sous forme de machos en direction de la galaxie m31 , phd thesis , paris magnier , e. a. , et al . 1993 , a&a , 272 , 695 lasserre , t. , et al . 2000 , a&a , 355 , l39 modjaz , m. , & li , w. d. 1999 , iau circ .
, 7218 , 2 paczyski , b. 1986 , apj , 304 , 1 riffeser , a. , et al .
2001 , astro - ph/0104283 udalski , a. , et al .
1993 , acta astronomica , 43 , 289 tomaney , a. , & crotts , a. p. 1994 , baas , 185 , # 17.01 witt , h. j. , & mao , s. 1994 , apj , 430 , 505
|
we present the first results of the analysis of data collected during the 1998 - 99 observational campaign at the 1.3 meter mcgraw - hill telescope , towards the andromeda galaxy ( m31 ) , aimed to the detection of gravitational microlensing effects as a probe of the presence of dark matter in our and in m31 halo .
the analysis is performed using the _ pixel lensing _ technique , which consists in the study of flux variations of unresolved sources and has been proposed and implemented by the agape collaboration .
we carry out a shape analysis by demanding that the detected flux variations be achromatic and compatible with a paczyski light curve .
we apply the durbin - watson hypothesis test to the residuals .
furthermore , we consider the background of variables sources .
finally five candidate microlensing events emerge from our selection . comparing with the predictions of a monte carlo simulation , assuming a standard spherical model for the m31 and galactic haloes , and typical values for the macho mass
, we find that our events are only marginally consistent with the distribution of observable parameters predicted by the simulation .
|
gravitational lensing signatures are a response to the projected surface mass column density of matter @xmath12 along the line of sight in galaxies and clusters . upon a suitable deprojection and circular averaging , an estimate of @xmath13 can ,
in principle , be used to trace the spherically averaged 3d density profile @xmath14 . in the past few years
, n - body simulations have shown [ @xcite , @xcite ( nav04 ) , @xcite ( m06 ) , @xcite(s09 ) ] that three - parameter models , especially the einasto @xcite profile and the @xcite de - projected sersic profiles ( ps97 ) , are able to produce better fits to the 3d density profiles of galaxy and cluster - sized dark matter haloes than two - parameter models ( @xcite @xmath15 ( 1996 ) ( nfw ) , @xcite ) .
while the ps97 profile has a well known 2d sky projected form - the sersic @xcite profile , there has been no such analytical counterpart for the einasto profile . in this paper
, we present a very good approximation for the 2d projection of the einasto family of 3d profiles .
thus , if the 3d total mass density is believed to be einasto - like , our model can be used to parametrically describe the projected 2d surface mass densities of galaxies and clusters in the weak and strong lensing regimes .
however , note that even upon radial averaging to smooth out the substructure , not all haloes subscribe to an einasto profile . for the rest of this paper
, we will limit the discussion to the 2d projection of the special case where the 3d profile is einasto - like . in 3d , the functional form of the einasto @xcite profile
is given by : @xmath16= -b[(\frac{r}{r_s})^{\frac{1}{n } } -1 ] \end{aligned}\ ] ] where , @xmath14 is the 3d ( spatial ) density at @xmath17 , @xmath18 ( or @xmath19 ) is the shape parameter , @xmath20 is a function of @xmath18 , @xmath21 the spatial density at a scale radius @xmath22 and @xmath23 . in 2d ,
the sersic @xcite profile , which has been used to describe the projected surface brightness profiles of galaxies , is similar in form to the einasto profile in 3d and is given by : @xmath24 = -q[(\frac{r}{r_e})^{\frac{1}{m } } -1 ] \end{aligned}\ ] ] such that @xmath25@xmath26@xmath27 where , @xmath28 is the projected distance in the plane of the sky , @xmath29 is the line of sight projected surface brightness at a projected scale radius @xmath30 , which can be defined to be the half - light radius of a sersic profile under the condition : @xmath31 ( ps97 ) with @xmath32 ( or @xmath33@xmath26@xmath34 ) characterizing the shape of the sersic profile .
the parameter @xmath4@xmath26@xmath35 in defines the shape of the einasto profile . in
the first ever fits to n - body haloes with a einasto profile , nav04 found an average value of @xmath36@xmath37@xmath38 for a wide range of halo masses from dwarfs to clusters . for galaxy - sized haloes @xcite found @xmath39@xmath3@xmath4@xmath3@xmath40 .
@xcite observed an evolution of @xmath4 with mass and redshift ( @xmath41 ) in the millennium simulation ( @xmath42 ) of @xcite and found @xmath4@xmath43@xmath44 for galaxy and @xmath43 @xmath45 for cluster - sized haloes .
a similar trend is supported by @xcite where @xmath4@xmath43@xmath9 for the most massive clusters in @xmath42 .
hence , although in this paper we discuss our approximation to a projected einasto profile for @xmath2@xmath3@xmath4@xmath3@xmath5 , particular attention is drawn to the domain @xmath4@xmath46@xmath9 ; where as we shall show in 3 , the errors due to our approximation out to @xmath47@xmath48 ( @xmath43@xmath7@xmath8 ) are @xmath49 @xmath50 .
the parameters @xmath22 , @xmath20 and @xmath18 in the einasto profile are not independent .
it can be seen , that in terms of a dimensionless length @xmath51 , the logarithmic slope of the density profile is given by : @xmath52 nav04 chooses to define @xmath22 such that @xmath53 ( the isothermal value of @xmath54 ) at @xmath55 and hence label @xmath22 as @xmath48 and @xmath21 as @xmath56 .
another approach is to use the convention of m06 , requiring @xmath22 to include half the total mass .
they quote a numerical estimate of @xmath20@xmath26@xmath57 . in this paper
we will follow the nav04 parametrization .
since the sersic profile describes the 2d surface brightness profiles of galaxies reasonably well , a natural question is : does the 2d surface mass density of einasto - like 3d dark matter haloes also follow a sersic - like description ? for the nav04 simulations , @xcite ( m05 ) have shown that a sersic function does produce fits with acceptable errors in the range of @xmath58 to @xmath8 , where @xmath58 is the minimum radius of convergence and @xmath8 is the virial radius in the n - body simulations of p03 and nav04 . in this paper , we focus on the analytical description of the projected einasto profile . in order to see
how well a projected einasto profile is described by a sersic profile , we numerically project and find that a sersic profile produces acceptable fits only in a limited range of the projected radius @xmath28 .
we shall show in 3 that this range also depends on whether one fits the numerically projected einasto profile in linear scale ( large errors with increasing r ) or log scale ( large errors with decreasing r ) of density .
it is thus evident that the sersic profile does not give an adequate description of a projected einasto profile for all @xmath28 .
the choice of sersic fit ( log or linear ) may , for example , have possibly strong implications in the strong and weak lensing regimes respectively , yielding incorrect results .
it is with this perspective , and the observation that the projection of an einasto profile ( i.e. integral of a sersic - like function ) is not a sersic - like function but rather a gamma - like function , we present a derivation of the surface mass density ( 2.1 ) for the einasto profile . throughout this paper
we will describe 3d spatial parameters as @xmath17 , @xmath22 ( or @xmath48 ) , @xmath21 ( or @xmath56 ) , @xmath20 and @xmath18 ( or @xmath59))and 2d projected parameters as @xmath28 , @xmath30 , @xmath60 and @xmath32 ( or @xmath61 ) .
we will also refer to the sersic profile as @xmath62 , our approximation to a projected 2d einasto profile as @xmath63 and a numerically projected einasto profile as @xmath64 .
further , for this paper , @xmath58 and @xmath8 have no physical meaning as such .
we note from the nav04 simulations , that on an average @xmath65 per cent of @xmath48 , and @xmath66 per cent of @xmath48 for galaxies and @xmath67 per cent of @xmath48 for clusters . in this paper
we will , for the sake of discussion , refer to @xmath68 @xmath69 @xmath48 and @xmath8 as @xmath43 @xmath70 @xmath48 .
in 2 we derive a semi - analytical form for @xmath63 , which for the nav04 parametrization of @xmath71 , simplifies to . in 3 we discuss details of estimating the fit parameters in our expression for @xmath63 , followed by a discussion of errors in our approximation .
we then present a comparison between @xmath64 and the best - fitting sersic profile @xmath62 . in
3.4 , we note the remarkable accuracy with which , along with and , can be used to extract the 3d parameters of einasto - like systems from its 2d form .
the einasto profile has generated considerable interest of late and has been used in recent studies of dark matter haloes ( nav04 , pr06 , m06 , g08 , hw08 , s09 ) . for the n - body haloes in nav04 ,
@xcite have shown that within the limited radial range of @xmath72 and shape parameter @xmath73 , a deprojected sersic profile also fits the 3d distributions almost as well as the einasto profile , and a sersic profile fits the non - parametrically estimated 2d surface densities of dark matter haloes with acceptable errors ( @xmath74 per cent ) . in the following discussion , we argue that if a 3d distribution is einasto - like , the 2d distribution need not be sersic - like and provide an analytically motivated functional form for the 2d projection , which can be used to describe the surface mass density of the einasto family of dark matter haloes subscribing to a wide range of the shape parameter ( @xmath75 or @xmath76 ) and over a wide radial ( projected ) range @xmath77 @xmath8 . in terms of the line of sight distance ( @xmath41 ) , the surface mass density @xmath78(r ) can be estimated from : @xmath79 while an exact analytical expression for the integral in for the einasto profile has so far eluded us , we derive below an excellent semi - analytical approximation . to intuitively motivate the functional form of @xmath80 for the einasto profile ,
observe that at @xmath81 , the integral presents us with an exact solution : @xmath82 \end{aligned}\ ] ] where , @xmath83 $ ] is the complete gamma function . hence , for r@xmath840
, it is reasonable to expect the integral to depend on terms involving incomplete gamma functions .
in fact , for the sake of discussion , one can make a very crude assumption that most of the contribution to the integral in at a given @xmath28 ( especially for @xmath85 ) comes from the region @xmath86 .
integrating , from some @xmath87r to @xmath88 ( where @xmath89 ) , one gets : @xmath90 \end{aligned}\ ] ] similarly , the integral for @xmath91 will have dominant contributions from terms involving @xmath92 $ ] ; where @xmath93 $ ] and @xmath94 $ ] are the upper - incomplete and lower - incomplete gamma functions respectively .
this is quite unlike a sersic function .
hence , although a sersic profile may fit a projected einasto profile in a limited range of @xmath28 it need not be a very good fit for all @xmath28 . with the 3d spatial distance @xmath95 , where @xmath28 is the 2d projected distance in the plane of the sky , and @xmath41 the line of sight distance from the object to the observer , at any given @xmath28 , one can define 3 regions for the integral in for the einasto profile : + region i : @xmath96 , integrating from @xmath97 to @xmath98 , with @xmath99 + region ii : @xmath100 , integrating from z=@xmath101 to @xmath88 , with @xmath102 , and + region iii : @xmath103 , in a neighborhood @xmath104 between @xmath98 and @xmath101 + + in region i : @xmath105 , the first term on the right hand side ( rhs ) of can be written as : @xmath106^{\frac{1}{2n } } \end{aligned}\ ] ] neglecting 4th and higher order terms in ( @xmath107 ) , in the binomial expansion of , the integral of ( from @xmath97 to @xmath108r ) has an analytical approximation : @xmath109^{\frac{1}{n } } -1\right)\right]\sqrt{\frac{2n}{b}}\left(\frac{r}{r_s}\right)^{\left(1-\frac{1}{2n}\right ) } \notag \\ \times \gamma\left[\frac{1}{2},\zeta^2_1\frac{b}{2n}\left(\frac{r}{r_s}\right)^{\frac{1}{n}}\right ] \end{aligned}\ ] ] in region ii : @xmath86 , a binomial expansion of the first term of the rhs of gives us : @xmath110 \end{aligned}\ ] ] fits to n - body simulations with the einasto profile have so far indicated an @xmath18@xmath84@xmath111 ( @xmath4@xmath46@xmath9 ) .
hence , observing that the leading contribution comes from the 1st term @xmath112 , one can drop the remaining terms .
this is especially true in our primary domain of interest ( @xmath113@xmath3@xmath18@xmath114 , i.e. @xmath4@xmath3@xmath9 ) . even for @xmath115@xmath18@xmath116 , although the approximation is not as good ( as it is for @xmath18@xmath117@xmath113 ) , it continues to provide better fits than a sersic profile in the entire range @xmath5@xmath3@xmath18@xmath3@xmath118 . with this understanding , the integral from @xmath119 to @xmath88 , in region
ii can be written as : @xmath120 \end{aligned}\ ] ] in region iii : @xmath121 , one can not make the approximations made in regions i and ii .
however , there exists a point @xmath122 between @xmath98 and @xmath101 , where the mean - value approximation will be valid in a domain @xmath104 about @xmath122 .
since the density profile falls rapidly for @xmath123 and gradually for @xmath124 , the domain of applicability of the mean - value approximation will be such that @xmath125 for @xmath85 ( @xmath126 as @xmath127 ) and tending to @xmath128 for @xmath124 .
further , it should be obvious that @xmath129 and @xmath104 will depend on the shape parameter @xmath18 as well .
a function describing @xmath104 , with such a property is : @xmath130 \end{aligned}\ ] ] with @xmath131 the remaining ( small ) excluded region does not add significantly to the integral ( refer to error plots in 3 ) . with this , the contribution to the integral in , from region iii- a domain @xmath104 around r - from an application of the mean - value theorem at @xmath122 can be written as : @xmath132\right ) \end{aligned}\ ] ] a few important observations are in order .
first , neglecting terms in the integrand of region i and ii , leads to over - estimating the integrands in those regions .
second , ignoring the contribution from region iii and fitting only for @xmath108 and @xmath133 , produces good fits ( refer to discussion following ) but understandably with a @xmath133 greater than @xmath133 with region iii included .
@xmath134 including region iii , lowers @xmath133 allowing it to be closer to 1 , resulting in a larger contribution from the upper - incomplete gamma function . a simultaneous fit of @xmath135 , @xmath136 and @xmath137(in @xmath104 ) accounts for these excess contributions through a negative sign from region iii . in region
i , since @xmath138 and higher order terms in @xmath107 are neglected , one can fix @xmath139 . further , although @xmath136 should in principle be estimated , we found it to be a reasonable approximation to fix @xmath140 .
this reduces the number of parameters to fit to only two - @xmath133 and @xmath137 - which in turn , during the fitting process , compensates for the approximations made on @xmath108 and @xmath136 . with these approximations
, the surface mass density @xmath63(r ) , for @xmath141 with @xmath142 can be written as : @xmath143 \notag \\ & \!+\frac{b^n}{2}\sqrt{\frac{2n}{b } } x^{\left(1-\frac{1}{2n}\right)}\gamma\left[\frac{1}{2 } , \zeta^2_1\frac{b}{2n } x^{\frac{1}{n}}\right ] e^{-b x^{\frac{1}{n } } } \\ & \!-\delta b^n x e^{-b\left(\sqrt{1+\epsilon^2}x\right)^{\frac{1}{n } } } \bigg ] \notag \end{aligned}\ ] ] where , @xmath144 for the nav04 parametrization of @xmath145 and by choosing @xmath146 , simplifies to ; where , after substituting factors like @xmath147 with @xmath5 in , we leave @xmath20 in the rest of the equation in order to reduce clutter .
this gives us : @xmath148 \ ! \notag \\
+ \frac{b^n}{2 } x^{\left(1-\frac{1}{2n}\right)}\gamma\left[\frac{1}{2 } , x^{\frac{1}{n}}\right ] e^{-b x^{\frac{1}{n } } } \!-\delta b^n x e^{-b\left(\sqrt{1+\epsilon^2}x\right)^{\frac{1}{n } } } \bigg ] \end{aligned}\ ] ] with , @xmath149\ ] ] @xmath150 @xmath151 @xmath152 @xmath153 and , @xmath133 and @xmath137 are numerically estimated ( fig.1 and fig.2 ) for the specific conditions of @xmath71 and @xmath154 to yield : @xmath155 @xmath156 we would also like to note that based on the accuracy of an approximation needed , one can neglect the contribution from region iii ( the @xmath104 term ) and with @xmath139 fit only for @xmath133 .
although we have not explored a functional relation of @xmath157 , we note that the best - fitting @xmath133 varies weakly from @xmath158 at @xmath159 to @xmath160 at @xmath161 to @xmath162 at @xmath163 , with a maximum error within @xmath164 ( usually reaching a peak around @xmath8 ) of @xmath165 at @xmath159 , @xmath166 at @xmath167 and @xmath168 at @xmath169 .
hence , within the current domain of @xmath170 from n - body simulations ( corresponding to @xmath171 ) one can neglect the @xmath104 term and set @xmath146 , and @xmath172 at the cost of an error in the range ( @xmath173 to @xmath174 ) .
the parametrizations and worked very well even at n=14 and n=0.95 .
however , we have not tested @xmath175 and @xmath176 values rigorously .
if greater accuracy is needed , we recommend fitting for @xmath133 and @xmath137.[refer to the discussion following , on why the @xmath104 term is negative ] . for @xmath177 ( @xmath178 ) .
each point has been obtained by fitting the numerically projected einasto profile with for @xmath71 and @xmath154.,width=317 ] @xmath177 ( @xmath178 ) .
each point has been obtained by fitting the numerically projected einasto profile with for @xmath71 and @xmath154.,width=317 ] this is a useful result because the surface mass density is expressed entirely as a function of the 3d spatial density parameters , which is not the case for any existing projected fitting function . equation or thus serves two purposes : one , given a 3d einasto profile , gives a good approximation to its 2d surface mass density . and two , a good fit to some 2d observations with for example surface mass density from lensing , will give us the 3d spatial density parameters of a einasto - like profile .
this is also quite unlike fitting 2d einasto profiles with sersic - like functions , where one first needs to fit for the 2-d and then deproject to fit for the 3d shape parameters ( m05 ) , which usually are different without any known existing functional relation between them .
we numerically integrate for the profile in and obtain @xmath179 in the domain @xmath180 @xmath48 for 90 profiles with a shape parameter in the range @xmath181 . a resolution in @xmath28 of @xmath182 @xmath48 ( @xmath43 @xmath69 @xmath58 ) ,
allows us to quantify errors due to our approximation in a domain @xmath183 , and we report comparison of errors up to @xmath47 @xmath48 or up to a r where @xmath184 whichever is earlier . with @xmath71 and @xmath154 , we use a non - linear least squares levenberg - marquardt algorithm to estimate the best - fitting values for @xmath133 and @xmath137 by fitting to each of the numerically generated @xmath64 profiles .
we find that @xmath133 and @xmath137 are best described by a second and third degree polynomial respectively , in @xmath59 ( fig.1 and fig.2 ) .
3 and fig . 4
describe the fractional error profile between our model @xmath63 and the numerically projected einasto profile @xmath64(r ) for a wide range of the shape parameter @xmath185 .
3 is more relevant to current n - body simulations , where in @xmath4 seems to be in the range 0.1 to 0.25 .
it is worth noting here , that for @xmath4 as high as 0.25 ( @xmath186 ) , the largest errors are @xmath187 in the range ( 0 to 30 ) @xmath48 . and
@xmath64 for @xmath188 , @xmath189 .
the x - axis plotted in log scale is expressed as a ratio of the 2d projected radius @xmath28 and the 3d scale radius @xmath48 of the corresponding einasto profile . @xmath190
@xmath48 while @xmath191 @xmath48.,width=317 ] and @xmath64 for @xmath192 , @xmath193 .
the x - axis plotted in log scale is expressed as a ratio of the 2d projected radius @xmath28 and the 3d scale radius @xmath48 of the corresponding einasto profile . @xmath190
@xmath48 while @xmath191 @xmath48.,width=317 ] in case
the range @xmath193 becomes relevant in the future , where n - body haloes for @xmath194 have not yet been found , we also present in fig .
4 a comparison between @xmath63 and @xmath64 .
this is also the domain where our assumptions in the @xmath86 region are weaker .
nevertheless , the accuracy of the approximation is striking with the worst error @xmath195 within @xmath196 @xmath48 . in this section
we discuss results of fits to @xmath64 with a sersic function and superimpose the @xmath63 model ( black solid lines ) for a relative comparison between @xmath62 and @xmath63 ( fig . 5 and fig .
the fits presented here were obtained using log scale of density .
one can also obtain fits with density in linear scale ; a comparison of residuals for the case @xmath197 is presented in fig .
the best fit parameters and consequently the error profile are quite different and is an indication that the best - fitting sersic profile does not provide an adequate representation of a projected einasto profile .
not evident in the plots ( fig.5 - 7 ) are the extremely large errors in the central density @xmath25 when fitting using log density . for @xmath198 ( fig .
5 ) the relative error at @xmath81 is @xmath199 , for @xmath197 ( fig.5 and 7 ) , the relative error is @xmath200 and @xmath201 for @xmath202 .
however , we observed empirically that fits in log density happen to reflect the significance of @xmath30 as enclosing half the total mass of @xmath64 ( it is to be noted that @xmath30 by definition encloses half the total mass of @xmath62 )
. this is because the domain over which @xmath62 ( with fits obtained in log density ) overestimates the density is a relatively small contributor to the total mass and fits with log density in the region @xmath203 are good . since , we can reliably estimate one of the sersic profile parameters ( @xmath30 ) through fits with log density ( as opposed to none in linear scale ) , we have presented results of fits in log density in fig
. 5 and fig.6 . the large errors in the central density also have serious consequences for strong lensing .
typically the strong lensing regime extends up to @xmath204 @xmath48 . in strong lensing
, image positions correspond to extremum ( minima , maxima and saddle ) points of the time delay surface .
the @xmath205 image @xmath206 for the @xmath207 source @xmath208 is given by : @xmath209 where @xmath210 is the normalized surface mass density at an angular position @xmath211 .
consequently contributions to the integral in from unusually large density near the center ( as a result of fits with a sersic profile in log density ) , will produce image separations larger than what one can expect from a numerical projection of the 3d einasto profile .
sersic profile fits to model a projected einasto profile should thus be avoided especially if the central region ( @xmath28@xmath49@xmath2@xmath48 ) is being excluded .
this is because , as shown in fig.7 , one can get a reasonably good fit ( relative error within @xmath212 ) over a large range of @xmath28@xmath213@xmath2@xmath48 ( fitting in log density ) giving an indication that the sersic profile is a good representation of projected einasto profile , but doing so will lead to even larger errors in the excluded central region ( @xmath28@xmath49@xmath2@xmath48 ) .
one should thus use caution in interpreting the other structural parameters , the shape parameter @xmath32 or @xmath33 and the central density @xmath25 .
fits in linear density present a different problem .
even though the central errors are much better ( relative error @xmath43@xmath212 for @xmath4@xmath26@xmath44 , fig.7 ) than fits using log density , the best - fitting @xmath30 does not enclose half the total mass of @xmath64 and the errors for large r keep increasing with r. although the sersic profile is not a good representation of a projected einasto profile , one can fit sersic profiles in limited domains of r and obtain an estimate of the shape of the projected einasto profiles in those domains only , but be careful to not use the resulting best - fitting values of @xmath30 and @xmath25 as a true representation of the half - mass radius and central density of the projected einasto . in fig.8 , we present two such relations between the 2d sersic index ( @xmath33@xmath26@xmath34 ) and the 3d einasto index @xmath4@xmath26@xmath35 in two domains @xmath28@xmath214 and @xmath28@xmath215 . in these 2 regions , @xmath33(@xmath4 )
can be described as power laws . for the domain @xmath28@xmath214
we find : @xmath216 and for @xmath203 defined as in 3.1 : @xmath217 not only is the shape ( @xmath33 ) different in the two domains , their evolution with @xmath4 is also different .
nevertheless , this result can be useful in obtaining an estimate of the shape of projected einasto profile in these two domains demarcated by @xmath48 .
the @xmath63 model , with errors @xmath218 does not face any of the above issues .
further , unlike the sersic profile , the @xmath63 model can predict the central density with almost @xmath219 errors due to the existence of an analytical solution .
thus , if the underlying 3d distribution is einasto - like , the 2d distribution should be modeled with @xmath63 . and @xmath64 with einasto index @xmath220 , @xmath189 with fractional error for the best - fitting @xmath63 ( black solid lines @xmath221 ) superimposed .
the x - axis plotted in log scale is expressed as a ratio of the 2d projected radius @xmath28 and the 3d scale radius @xmath48 of the corresponding einasto profile . @xmath190
@xmath48 while @xmath191 @xmath48.,width=317 ] and @xmath64 with einasto index @xmath222 , @xmath193 with fractional error for best - fitting @xmath63 ( black solid lines @xmath221 ) superimposed .
the x - axis plotted in log scale is expressed as a ratio of the 2d projected radius @xmath28 and the 3d scale radius @xmath48 of the corresponding einasto profile . @xmath190
@xmath48 while @xmath191 @xmath48.,width=317 ] from fits to @xmath64 ( for @xmath197 ) with a sersic profile using log density ( dot - dashed line ) and a sersic profile using linear density ( dashed line ) .
not shown in the plots are the errors in central density @xmath223 .
the errors are @xmath200 for the fit with log density and @xmath212 for the fit with linear density .
the fractional error in @xmath63 ( black solid line @xmath221 ) is superimposed for comparison .
the x - axis , plotted in log scale , is a ratio of the 2d projected radius @xmath28 to the 3d scale radius @xmath48 of the corresponding einasto profile . @xmath190
@xmath48 while @xmath191 @xmath48.,width=317 ] = 1/m ) of a projected ( 2d ) einasto profile as a function of the 3d einasto index ( @xmath4=1/n ) in two domains @xmath224 ( dotted ) and @xmath203 ( solid ) indicating the absence of a unique sersic index @xmath33 for all r. the power law relations for @xmath33(@xmath4 ) in equations and are applicable only in these domains.,width=317 ] the @xmath63 model is not just a good description of the projected einasto profile but is also expressed in terms of the 3d einasto profile parameters .
it should thus be possible to recover the 3d parameters ( @xmath225 ) from fits to 2d distributions that subscribe to an underlying 3d einasto - like system .
for the wide family of numerically projected einasto profiles @xmath64 described in this paper , we could recover the 3d parameters for all of them with an accuracy of @xmath226 or better , by fitting @xmath64 with and the parametrizations of and through a non - linear least squares levenberg - marquardt algorithm .
given that , as of now robust data ( within virialized regions ) from n - body simulations are in the domain @xmath58 to @xmath8 , the fits were performed in this domain . in passing ,
we note that our results are even better if we fit from @xmath6 to @xmath47 @xmath48 .
such a high degree of accuracy indicates that if the 2d distribution is indeed a projected einasto profile , the 3d parameters can be recovered very well even from a limited radial range of observations .
we note that , for the entire range of @xmath227 the fits always converged for an intial guess of @xmath228 @xmath4 , @xmath229 @xmath230 and @xmath56 in the range ( @xmath2 to @xmath231 ) @xmath232 .
an inital guess of very low @xmath233 and a guess for @xmath234 for the type of object ( galaxy or cluster ) being considered can be a reasonable starting value for the fit to converge .
we also did not encounter any local minima .
i.e. if the fit converges , it always converged to the true set of ( @xmath4 , @xmath48 , @xmath56 ) .
non - parameteric estimates of density profiles in n - body simulations ( nav04,m06 ) favour einasto - like profiles , since they provide better fits than the two - parameter nfw and moore profiles .
@xcite have also shown that a de - projected sersic profile fits the 3d halo mass distribution almost as well as the einasto profile , and a sersic profile provides good fits to non - parametric estimates of surface mass densities ( m05 ) of the nav04 n - body haloes .
we have observed that fits with a sersic function ( @xmath62 ) to a numerically projected einasto profile ( @xmath64 ) are sensitive to whether one fits using linear density @xmath62 ( errors increasing for large r ) or log density @xmath235 ( errors increasing for small r ) yielding widely varying results .
consequently , the sersic profile does not give an adequate description of the projected einasto profile .
sersic profile fits to the surface mass density of n - body haloes ( m05 ) , whose 3d spatial densities are well fit by einasto profiles with @xmath236@xmath3@xmath4@xmath3@xmath237 , have been obtained from the limited radial range of @xmath58 to @xmath8 . for the haloes in m05 and nav04 ,
this range is generally less than two decades in radius .
hence , if the 3d distribution is indeed einasto - like , interpreting structural properties from fits with a sersic profile , especially @xmath32 in as the shape parameter , @xmath25 as the central density and @xmath30 as the half - mass radius , can be misleading . in this paper
, we have provided an analytical approximation to the projected surface mass density of einasto - like 3d density distributions .
the fit errors are well contained to @xmath49@xmath238 for the projected radial range @xmath6@xmath3@xmath28@xmath239 @xmath48 equivalent to @xmath7@xmath8 and shape parameter , @xmath5@xmath3@xmath18@xmath3@xmath118 , or @xmath2@xmath3@xmath4@xmath3@xmath5 .
this model can therefore be used both as a fitting function for 2d observations and also to extract the 3d parameters of einasto - like profiles .
since @xmath63 fits a projected einasto profile in a wide radial range , it can be used for fitting strong and weak lensing observations in systems whose total 3d density distribution is believed to be einasto - like .
one can also numerically integrate to get reliable estimates of the mass enclosed .
finally , we note that the form similarity of and , i.e. fitting functions that describe the 3d mass density of dark matter haloes and the 2d light distributions of galaxies , respectively , could be largely coincidental and should be used with caution when drawing conclusions about the similarity of dynamical evolution that lead to the formation of the stellar components of ellipticals and dark matter haloes .
bkd and llrw would like to acknowledge the support of nasa astrophysics theory grant nnx07ag86 g .
we thank jaan einasto and urmas haud for pointing us to the original literature on the einasto profile .
50 einasto j. , 1965 , tartu astron .
teated , no.17 , 1 gao l. , navarro j.f . , cole s. , frenk c.s . ,
white s.d.m . , springel v. , jenkins a. , neto a.f . , 2008 , mnras , 387,536 ( g08 ) hayashi e. , white s.d.m . , 2008 , mnras , 388,2 ( hw08 ) merritt d. , navarro j.f . , ludlow a. , jenkins a. , 2005 , apj , 624:l85 ( m05 ) merritt d. , graham a.w . , moore b. , diemand j. , terzic b. , 2006 , aj , 132:2685 ( m06 ) moore b. , quinn t. , governato f. , stadel j. , lake g. , 1999 , mnras , 310 , 1147 ( moore 99 ) navarro j.f .
, frenk c.s . ,
white s.d.m . , 1996 , apj , 462 , 563 ( nfw ) navarro j.f . , frenk c.s . , white s.d.m . , 1997 ,
apj , 490 , 493 ( nfw ) navarro j.f .
et al . , 2004 ,
mnras , 349 , 1039 ( nav04 ) power c. , navarro j.f . ,
jenkins , a. , frenk c.s .
, white s.d.m . , springel v. , stadel j. , quinn t. , 2003 , mnras , 338 , 14 ( p03 ) prada f. , klypin a.a . , simonneau e. , betancort - rijo j. , patiri s. , gottlober s. , sanchez - conde m.a . , 2006 , apj , 645:1001 ( pr06 ) prugniel ph . and simien f. , 1997 , aa , 321 , 111 ( ps97 ) sersic j.l .
1968 , atlas de galaxias australes , observatorio astronomico de cordoba , cordoba , argentina springel v. et.al .
2005 , nature , 435,629 stadel j. , potter d. , moore b. , diemand j. , madau p. , zemp m. , kuhlen m. , quilis v. , 2009 , mnras , 398,1,l21 ( s09 )
|
recent advances in n - body simulations of dark matter haloes have shown that three - parameter models , in particular the einasto profile characterized by @xmath0 with a shape parameter @xmath1 , are able to produce better fits to the 3d spatial density profiles than two - parameter models like the navarro , frenk and white ( nfw ) , and moore et al . profiles . in this paper , we present for the first time an analytically motivated form for the 2d surface mass density of the einasto family of dark matter haloes , in terms of the 3d spatial density parameters for a wide range of the shape parameter @xmath2 @xmath3 @xmath4 @xmath3 @xmath5 .
our model describes a projected ( 2d ) einasto profile remarkably well between @xmath6 and @xmath7 @xmath8 , with errors less than @xmath9 per cent for @xmath4 @xmath10 and less than @xmath11 per cent for @xmath4 as large as 1 .
this model ( in 2d ) can thus be used to fit strong and weak lensing observations of galaxies and clusters whose total spatial(3d ) density distributions are believed to be einasto - like .
further , given the dependence of our model on the 3d parameters , one can reliably estimate structural parameters of the spatial ( 3d ) density from 2d observations .
we also consider a sersic - like parametrization for the above family of projected einasto profiles and observe that fits with a sersic profile are sensitive to whether one fits the projected density in linear scale or logarithmic scale and yield widely varying results .
structural parameters of einasto - like systems , inferred from fits with a sersic profile , should be used with caution .
[ firstpage ] gravitational lensing galaxies : clusters : general galaxies : fundamental parameters galaxies : haloes galaxies : structure dark matter .
|
recently , euclidean simplicial quantum gravity(sqg ) with the additional matter fields has been investigated with using the monte - carlo simulation . from the previous numerical simulation@xcite
, we expect the four - dimensional(4d ) simplicial model realizes the realistic 4d quantum gravity .
especially , in ref .
@xcite , it is shown that the relation between sqg and the 4d conformal gravity@xcite , calculating with the grand - canonical method . however , in order to discuss the relation between the discretized model and the continuum theory in detail , we need to analyze the correlations on the 4d simplicial manifold @xmath1 .
namely , we need to discuss the geometry on @xmath1 instead of the statistical properties which have been discussed until now . in order to discuss the geometry on @xmath1
, we analyze the configurations of the 4d sqg coupled with @xmath2 u(1 ) gauge fields and @xmath3 massless scalar fields . the partition function @xmath4 is given as the sum of the all of the configuration of 4-simplices @xmath5 , @xmath6 where @xmath7 , @xmath8 and @xmath9 is the action for the gravity , the scaler field @xmath10 on the vertex @xmath11 and the vector field @xmath12 on the link @xmath13 respectively .
for the action of the gravity part , we use the discretized einstein - hilbert action , @xmath14=\kappa_{4}n_{4}-\kappa_{2}n_{2}\ ] ] where @xmath15 denotes the number of the @xmath11-simplices , @xmath16 is related to newton s constant and @xmath17 is related to cosmological constant .
the action for the scalar field @xmath10 on the vertex @xmath11 becomes @xmath18 where @xmath19 is number of 4-simplices sharing link @xmath20 , then , the action of vector field @xmath21 on link @xmath20 @xmath22 where @xmath23 is number of 4-simplices sharing triangle @xmath24 . in order to attempt the canonical simulation
, we add the volume fixing term @xmath25 to the total action , @xmath26 where @xmath27 is the parameter to fix the volume @xmath28 and @xmath29 denotes the size of the target volume . at @xmath30 , we take a sample of @xmath1 and attempt to measure .
in order to analyze the geometry of @xmath1 , we discuss the fractal properties based on the geodesic distance with measuring the fractal dimension and the correlation functions of the additional fields .
we consider some different definitions of the geodesic distance on @xmath1 .
one of such the definitions is the shortest distance between two 4-simplices along the 4-simplices , so - called the simplex - simplex geodesic distance .
however , we expect that the expansion of the volume along the simplex - simplex geodesic distance is restricted , since the number of 4-simplices boarding on the 4-simplex is fixed .
another definition is the shortest distance between two vertices along the link , so - called the vertex - vertex geodesic distance . in order to discuss the difference between the two distances , we introduce the box - counting dimension @xmath31 on @xmath32 , @xmath33 where @xmath0 denotes the scale of the geodesic distance
, @xmath34 denotes the number of the closed - surface @xmath35 with the radius @xmath36 .
the box - counting method is known as one of the standard methods to define the fractal dimension . in figure
[ fig.bc ] , we show the @xmath37 based on each geodesic distances in the case of pure gravity .
the numerical results suggest that each geodesic distance need rescaling in order to be equivalent to the other definition of geodesic distance .
we consider that this rescaling factor is caused by the difference of the degree of freedom of each distances on the discretized manifold . from the behavior of the box - counting dimension in the branched polymer region ,
the simplex - simplex geodesic distance is the intrinsic scale and that it needs the rescaling factor for agreement with the unique shortest distance on the manifold . on the contrary
, numerical results suggest that the vertex - vertex geodesic distance does nt need the extra rescaling factor and that it is close to the unique shortest distance .
then , we choose the vertex - vertex geodesic distance for analysis of the geometry of @xmath38 .
we analyze the correlation functions on the vertex - vertex geodesic distance instead of the conventional geodesic distance of simplex - simplex distance . in order to analyze the behavior of the correlation function
, we introduce the correlation length @xmath39 for the operator @xmath40 on the vertex @xmath41 , @xmath42 this value diverges to infinity at the critical point of second order phase transition . in figure [ fig.col ]
, we show the numerical results in the case of the pure gravity and the additional one vector field @xmath43 .
the numerical result suggests that the value of @xmath39 indicates the property of manifold and the phase structure .
moreover , from the finite - size - scaling of the value of @xmath39 at the critical point between the crumpled phase and smooth phase in the case of @xmath43 , @xmath39 diverges to infinity for the large @xmath28 limit . while the divergence of @xmath39 is not found both at the critical point of the pure gravity case and at the obscure transition point between the smooth phase and the branched polymer .
the numerical results suggest that the type of transition at the critical point of the additional matter case is different from the type of both the critical point of the pure gravity and the obscure transition point . actually , we found the power low behavior of the correlation functions at the critical point in the case of @xmath43 . in figure
[ fig.tf_na=1 ] , we show the correlation functions for the local scalar curvature . the deviation from the power low can be considered as the finite size effect .
in order to discuss the geometry on @xmath1 , we analyze the geodesic distances defined along the lattice building blocks . from the numerical result of the box - counting dimension ,
the simplex - simplex geodesic distance need extra in order to be equivalent to the vertex - vertex geodesic distance .
moreover , we show the behavior of the correlation function numerically .
the correlation length realizes the phase diagram and its peak value diverges to the infinity at the critical point for the large @xmath28 limit .
then , the numerical results of the box - counting dimension and the correlation function suggest that the massless mode propagates on the simplicial manifold which has the fractal dimension nearly 4 . in conclusion
, the geometry on the simplicial manifold can be analyzed on the basis of the vertex - vertex geodesic distance .
|
the fractal properties of four - dimensional euclidean simplicial manifold generated by the dynamical triangulation are analyzed on the geodesic distance @xmath0 between two vertices instead of the usual scale between two simplices . in order to make more unambiguous measurement of the fractal dimension
, we employ a different approach from usual , by measuring the box - counting dimension which is computed by counting the number of spheres with the radius d within the manifold .
the numerical result is consistent to the result of the random walk model in the branched polymer region .
we also measure the box - counting dimension of the manifold with additional matter fields .
numerical results suggest that the fractal dimension takes value of slightly more than 4 near the critical point .
furthermore , we analyze the correlation functions as functions of the geodesic distance .
numerically , it is suggested that the fractal structure of four - dimensional simplicial manifold can be properly analyzed in terms of the distance between two vertices .
moreover , we show that the behavior of the correlation length regards the phase structure of 4d simplicial manifold .
|
spin properties of electrons in semiconductor heterostructures have become in recent years subject of intense experimental and theoretical interest because of their inherent scientific value as well as possible spintronic applications . among numerous heterostructures
, the system gaas / ga@xmath3al@xmath4as has won a unique position .
first , gaas is after silicon the most important semiconductor material .
second , the system gaas / ga@xmath3al@xmath4as has well established parameters which allows the theorists to describe new subtle phenomena .
third , due to the advanced growth technology the electrons in gaas quantum wells have very high mobilities making it possible to detect even weak effects .
thus it is of interest to describe precisely the spin and orbital electron energies in gaas / ga@xmath3al@xmath4as quantum wells not only for their own sake , but also as a model example for other heterostructures . from the theoretical point of view ,
a description of magnetooptical effects in gaas - type materials is not an easy task since gaas is a medium gap semiconductor . as a result , its band structure exhibits features characteristic of narrow - gap materials , but it is insufficient to treat them by the models normally used for such systems .
we showed in our previous work on bulk gaas and inp that the simplest adequate description of the band structure of these materials is given by the so called five - level * p@xmath0p * model [ 1 , 2 ] .
thus , the present work on the spin @xmath1-factor of electrons in gaas / ga@xmath3al@xmath4as quantum wells was first motivated by the fact that the theoretical description used in the literature for this purpose had been based on the three - level model [ 3 ] .
examining the problem we realized that the theory [ 3 ] suffers from other deficiencies , as we will demonstrate below .
the theory we develop for the spin energies is equally valid for the cyclotron ( orbital ) energies , so that we describe also the cyclotron masses , although here our treatment improves only slightly the existing approach [ 4 ] .
an important feature of the iii - v compounds is a bulk inversion asymmetry ( bia ) of these materials . as is known from the fifties [ 5 ] , the bia results in a spin splitting of energies for a given direction of the wavevector * k*. it is , however , of interest to investigate how this spin splitting behaves in the presence of an external magnetic field .
one can rephrase the problem by asking how the zeeman spin splitting caused by the magnetic field combines with the spin splitting caused by bia . in the bulk magneto - optical or magneto - transport studies one is usually interested in the landau levels at @xmath9 = 0 , since the density of landau states has singularities at the vanishing @xmath9 .
however , in a quantum well one deals with electric subbands for which the value of @xmath10 is `` frozen '' in the subband wavefunction along the growth direction .
the bia splitting is sensitive to this @xmath10 value which increases with a decreasing well width .
this in turn is reflected in the spin @xmath1-factor . in the following we
will be concerned with symmetric quantum wells , so that the bychkov - rashba spin splitting , caused by the structure inversion asymmetry , does not come into play [ 6 ] .
our calculation of the spin and cyclotron electron energies in gaas / ga@xmath3al@xmath4as heterostructures is based on the five - level * p@xmath0p * model ( 5lm ) .
an important advantage of the model is that it includes the bia mechanism of the spin splitting .
we solve the eigenenergy problem by an iteration procedure which is more precise than an expansion in powers of momentum used in the literature .
we check the precision of consecutive iteration steps .
the results are compared with those obtained by the three - level * p@xmath0p * model ( 3lm ) and it is shown that the latter is insufficient for the gaas - type materials . our paper is organized as follows . in section
ii the * p@xmath0p * theory within 5lm is formulated for the bulk semiconductor . next
the 3lm model is used to obtain results for the spin @xmath1-value .
it is demonstrated that the procedure of ref . [ 3 ] based on the same model is inadequate .
the iteration solution of 5lm is checked against `` exact '' results for the bulk , as obtained by a numerical procedure . in section
iii the theory for the quantum wells is worked out from the 5lm matrix .
results for the spin @xmath1-factor and the cyclotron mass are presented , compared with experimental data and discussed in section iv .
the paper is concluded by a summary .
in order to discuss the previous treatments and to establish standards of precision for various approximations we first consider the 3d case .
the * p@xmath0p * theory , which is the * k@xmath0p * theory generalized for the presence of an external magnetic field * b * , has the form ( cf .
[ 1 ] ) @xmath11 @xmath12f_l = 0\ ] ] where @xmath13 is the energy , @xmath14 is the kinetic momentum , @xmath15 is the vector potential of magnetic field @xmath16 , and @xmath17 . here
@xmath18 are the pauli spin matrices , @xmath19 is the volume of the unit cell , @xmath20 are periodic amplitudes of the luttinger - kohn functions , @xmath21 is the bohr magneton , @xmath22 are the interband matrix elements of momentum and @xmath23 are those of spin - orbit interaction .
the sum in eq .
( 1 ) runs over all bands @xmath24 included in the model , @xmath25 runs over the same bands , @xmath26 are the band - edge energies , see below . five - level model for the band structure of gaas - type semiconductors .
energy gaps , spin - orbit energies , interband matrix elements of momentum and of the spin - orbit interaction are indicated .
letters c and c mark symbolically far - band contributions to the effective mass and the spin @xmath1-factor of conduction electrons , respectively . ] in the following we shall be concerned with the five - level model of the band structure , as illustrated in fig
. 1 . within this model there exist three interband matrix elements of momentum @xmath27 @xmath28@xmath29 and three matrix elements of the spin - orbit interaction @xmath30_y|z > \;\;\;,\ ] ] @xmath31_y|z ' > \;\;\;,\ ] ] @xmath32_y|z ' > \;\;\;,\ ] ] where the nonprimed functions are related to the @xmath33 , @xmath34 valence bands , while the primed functions are related to the @xmath35 , @xmath36 conduction bands .
the band edge energies @xmath26 are generally influenced by @xmath37 ( cf .
1 ) but in ( 8) gaas - type materials , where @xmath37 is small , this feature may be ignored and we have : @xmath38 .
we take the zero of energy at the @xmath39 conduction band edge . within 5lm
there are 14 basis functions and the * p@xmath0p * differential hamiltonian for the envelope functions @xmath40 has the explicit form of eq .
we define @xmath41 , @xmath42 , @xmath43 , @xmath44 . here
@xmath45 is the cyclotron frequency with free electron mass .
the quantities @xmath46\pm c'\mu_bb$ ] result from the far band contributions to the conduction band energies .
the operators @xmath47 are proportional to the raising and lowering operators for the harmonic oscillator functions : @xmath48 and @xmath49 , in which @xmath50 is the magnetic radius .
the matrices @xmath51 , @xmath52 and @xmath53 have different forms for various directions of magnetic field . in the following
we shall be concerned with the case * * b**@xmath54[001 ] crystal direction , for which @xmath55\;\;,\ ] ] + @xmath56\;\;.\ ] ] for @xmath57=0 and @xmath9=0 matrix ( 8) factorizes into two 7@xmath587 matrices containing spin - up and spin - down electron energies . for @xmath57 = 0 and @xmath59 0
the resulting bands are spherical ( i.e. the energies do not depend on the direction of * b * ) , but the spin - up and spin - down states are mixed by the @xmath9 terms .
still , in this case one can find solutions of the eigenvalue problem in form of single harmonic oscillator functions . if @xmath600 , the resulting bands are nonspherical and the above simple solutions do not exist . at this point
we emphasize that matrix ( 8) contains no approximations within 5lm . for considerations of the bulk
there is no external potential . to describe the magnetic field
we take the gauge * * a**=[-by , 0 , 0 ] , corresponding to * * b**@xmath61[001 ] .
one can then look for the envelope functions in the form @xmath62 , where @xmath63 are the harmonic oscillator functions .
the three level model of @xmath39 , @xmath33 , @xmath34 levels ( see fig .
1 ) is not adequate for the description of the conduction band in gaas - type materials ( see [ 1 , 2 ] ) , but it can be solved exactly and we use it as a starting point of our considerations as well as for a discussion of the existing work on the @xmath1-values .
once the higher conduction levels are omitted , the couplings @xmath64 , @xmath57 and @xmath65 do not come into play ( see fig .
1 ) , the conduction band is spherical and the resulting 8@xmath588 hamiltonian may be solved in terms of eight harmonic oscillator functions .
this was first done by bowers and yafet [ 7 ] , see also [ 8 , 9 ] . in our more general formulation ( 8)
, the 3lm corresponds to considering the columns and rows 4 , 5 , 6 , 7 , 11 , 12 , 13 , 14 .
the resulting energy for the spin - up state of the n - th landau level is given by @xmath66+\ ] ] @xmath67+\ ] ] @xmath68\;\;,\ ] ] while the energy for the spin - down state of the n - th landau level is given by @xmath69+\ ] ] @xmath70+\ ] ] @xmath71\;\;,\ ] ] where @xmath72 @xmath73 @xmath74 @xmath75 @xmath76 @xmath77 the first terms in eqs .
( 11 ) and ( 12 ) represent the orbital parts , the second are related to the spin parts ( they differ in sign for spin - up and spin - down states ) and the third give @xmath78 parts for the motion along the magnetic field .
equations ( 11 ) and ( 12 ) represent fourth - order polynomials for the unknown energies . if we neglect the free - electron and far - band contributions , eqs .
( 11 ) and ( 12 ) reduce to cubic equations for the energies .
an influence of free electron terms on the description of spin @xmath1-factors in selected materials was discussed by singh et al [ 10 ] . in investigations of bulk semiconductors
one is usually interested in the simplified case of @xmath9 = 0 since the singular density of states for the landau levels corresponds to the vanishing @xmath9 .
however , our final aim in this work is to investigate quantum wells , which corresponds to the situation of @xmath59 0 .
solving eqs . ( 11 ) and ( 12 ) for the spin - up and spin - down energies one can find the cyclotron effective mass and the spin @xmath1-value exactly within 3lm .
we define the cyclotron mass @xmath2 and the spin @xmath1-value in the standard way @xmath79 @xmath80 the dependence of @xmath2 and @xmath1 on the electron energy was established in the early days , see refs [ 11 , 12 ] . knowing the solutions of eqs .
( 11 ) , ( 12 ) and fixing magnetic field intensity @xmath81 one can determine the dependence of @xmath2 and @xmath1 on @xmath78 exactly within 3lm .
since the conduction band in gaas is only weakly nonparabolic , we can expand the exact energies using eqs .
( 11 ) and ( 12 ) in the limit @xmath82 .
the zero - order terms give @xmath83 and @xmath84 values at the conduction band edge .
the first - order terms give the first nonparabolic approximation .
one can alternatively say that the zero - order terms are proportional to @xmath85 while the first - order terms are proportional to @xmath86 . using eqs .
( 11 ) , ( 12 ) and the definitions ( 14 ) and ( 15 ) we obtain after some manipulation ( we keep the free electron terms and the far - band contributions only in the conduction band ) @xmath87\;\;,\ ] ] @xmath88 @xmath89\;\;,\ ] ] where @xmath90 , and @xmath91 @xmath92 are the band - edge values of the effective mass and of the @xmath1-factor according to 3lm .
we define @xmath93 . to calculate the numerical values we take the following band parameters for gaas : @xmath94 = 27.865ev , @xmath95 = -1.519 ev , @xmath96 = -1.86 ev , @xmath97 = -3.070 , @xmath98 = -0.102 .
this results in @xmath83 = 0.0660 @xmath99 [ 2 ] and @xmath84 = -0.44 [ 13 ] .
we would like to plot the @xmath1 value as a function of energy in view of the applications to quantum wells ( qw ) .
to this end we calculate the theoretical energy @xmath100 for the ground electric subband in a rectangular ga@xmath5al@xmath6as / gaas / ga@xmath5al@xmath6as qw as a function of the well width @xmath101 .
next we define the value of @xmath102 by the equality : @xmath103 , where @xmath83 = 0.0660 @xmath99 .
this value of @xmath104 is introduced to the above relations for @xmath1 . when describing the final results for the @xmath1-value and the cyclotron mass in qw we will use the same range of well widths .
theoretical spin @xmath1-factor of conduction electrons in bulk gaas _ versus _ the wave vector @xmath102 parallel to the magnetic field * b * , or the corresponding well width @xmath101 ( see text ) .
the solid line ( ex ) shows exact results from the three - level model , the dashed line ( ap ) represents the first nonparabolic approximation to 3lm , the dashed - dotted line ( ik ) illustrates the results of ref .
[ 3 ] also based on 3lm ( see text ) . ] in fig .
2 we plot the spin @xmath1-value of the conduction electrons in bulk gaas as a function of @xmath101 ( or @xmath102 ) .
the solid line indicates the exact values of @xmath1 within 3lm , as obtained from eqs . ( 11 ) and ( 12 ) for @xmath81 = 2 t. the free - electron and distant band contributions are kept only for the conduction band .
the dashed line indicates values obtained from eq .
( 17 ) , resulting from the expansion for weak nonparabolicity . as can be expected , for bigger widths @xmath101 ( or smaller values of @xmath102 ) the approximated values approach the exact ones .
finally , the dashed line marked with ik indicates values obtained from eq .
( 9 ) of ref . [ 3 ] with the use of the same material parameters .
it can be seen that ref . [ 3 ] predicts a much stronger dependence of @xmath1 on @xmath102 than both our curves and , in particular , our dashed curve which has been calculated in the same nonparabolic approximation as that used in ref . [ 3 ] .
the reason of this discrepancy is that the formulas given in ref .
[ 3 ] correspond to our eq .
( 17 ) , with our @xmath105 replaced by @xmath106 and our @xmath84 replaced by @xmath107 . in case of the mass
the approximation used in ref .
[ 3 ] is not too bad since it misses only the term of 1+c [ cf .
( 18 ) ] . in case of the @xmath1-value , however , it is seriously inadequate since it misses the additive term of + 2 + 2c , which is essential for the band - edge value of @xmath84 in gaas , see eq .
( 19 ) . as a result , the second term in the square bracket of eq .
( 17 ) appears not with the coefficient @xmath84 = -0.44 , as it should , but with the coefficient of about -2.24 .
this gives a much too strong dependence of @xmath1 on @xmath108 , as illustrated in fig .
2 . the dependence of @xmath1 on @xmath101 ( or @xmath102 ) , as given by the exact solution of 3lm ( curve ex in fig .
2 ) , does not describe experimental data in gaas / ga@xmath5al@xmath6as quantum wells . in particular , it does not exhibit the change of sign to positive @xmath1-factors for small values of @xmath101 . on the other hand , we know that the inclusion of higher conduction @xmath33 , @xmath34 bands in the * p@xmath0p * description ( see fig .
1 ) results in stronger conduction band s nonparabolicity and , in particular , it gives the electron @xmath1-values that change sign to positive at higher electron energies [ 1 ] .
thus it is necessary to include the @xmath35 , @xmath36 levels in the refined description of @xmath1 in gaas .
this is equivalent to using the full matrix ( 8) for the determination of electron energies . in view of the applications to quantum wells
the @xmath9 terms must be retained .
we pursue this program below . in this section
we calculate the spin @xmath1-value of conduction electrons using the five - level * p@xmath0p * model .
this model is adequate for the gaas - type materials .
once the two upper conduction levels are included , the matrix elements @xmath64 , @xmath57 and @xmath109 come into play , see fig .
1 . as a result , two qualitatively new features appear .
first , the appearance of the matrix element @xmath57 makes the conduction band nonspherical .
second , the matrix element @xmath57 does not vanish if the crystal is characterized by the bulk inversion asymmetry ( bia ) .
this results in the spin splitting of conduction energies for a given direction of the wave vector * k * ( the dresselhaus mechanism [ 5 ] ) .
we develop a workable approach to the calculation of @xmath1-factor in quantum wells based of 5lm using an iterative expansion in powers of the interband matrix elements @xmath110 , @xmath64 and @xmath57 .
clearly , it is necessary to check whether such an expansion is adequate . in other words
, we should make sure that the expansion procedure gives the @xmath1-values comparable with those obtained from the complete matrix ( 8) . to this end we first solve the eigenenergy problem given by the * p@xmath0p * hamiltonian ( 8) by an alternative method , first used by evtuhov for the nonspherical valence bands of ge [ 14 ] . as already mentioned , for spherical energy bands the solutions of the eigenenergy problem in the presence of a magnetic field
are given in terms of simple harmonic oscillator functions .
for not too strong deviations from band s sphericity we look for the envelope functions in the form of sums of harmonic oscillator functions with unknown coefficients @xmath111 since our hamiltonian ( 8) contains raising and lowering operators for the harmonic oscillators , it is possible to perform the prescribed operations .
then we multiply the obtained eigenenergy equations on the left by consecutive harmonic oscillator functions and integrate the scalar products using the ortonormality relations .
this results in equations for the coefficients and the condition for nontrivial solutions gives the eigenenergies .
in other words , the evtuhov method is a standard way of transforming a differential eigenvalue problem into an algebraic problem by taking for the complete set of functions the harmonic oscillator functions .
not going into details of the procedure we will give a general scheme for the resulting eigenenergy matrix .
the elementary irreducible block is a 7@xmath587 matrix , one containing the spin - up state and another the spin - down state .
if @xmath57 and @xmath9 terms are neglected , the big matrix factorizes into the 7@xmath587 matrices for the spin - up and spin - down energies for consecutive landau levels . the nondiagonal @xmath57 and @xmath9 terms couple the elementary 7@xmath587 matrices .
we find that in order to obtain precise energies for the lowest landau levels we need to go to 70@xmath5870 matrices .
the schematic form of such a matrix is given in eq .
( 21 ) .
@xmath112\ ] ] we use a symbolic notation , for example @xmath113 stands for the elementary 7@xmath587 matrix containing the conduction landau level @xmath114 with spin up .
the symbols @xmath115 and @xmath116 stand for 7@xmath587 matrices containing @xmath117 , @xmath118 and @xmath116 terms , respectively , while @xmath119 stands for 7@xmath587 matrices containing @xmath119 terms . to calculate the @xmath1-value we will be interested in the lowest conduction levels @xmath120 . for @xmath121=0 , matrix ( 21 ) gives the energy of the conduction level @xmath122 ( second row and column ) perturbed by interactions with other levels .
for @xmath121=1 we obtain the energy of the @xmath123 level ( first raw and column ) . fixing the values of @xmath81 and @xmath9 we obtain from matrix ( 21 ) the energies which we consider to be `` exact ''
, so we use them as standards in the estimation of following expansions .
.band parameters of ga@xmath3al@xmath124as alloys for different chemical compositions @xmath125 , as used in the calculations , see [ 13 , 15 - 18 ] .
energies e@xmath126 , g@xmath126 , e@xmath127 , g@xmath127 are defined in fig . 1 , c and c are far - band contributions to the band - edge values of @xmath128 and @xmath84 , respectively , see eqs . ( 18 ) and ( 19 ) .
the interband matrix elements of momentum and of the spin - orbit interaction are taken to be independent of @xmath125 : @xmath94=27.865 ev , @xmath129=2.361 ev , @xmath130=15.563 ev , @xmath65=-0.061 ev . [ cols="^,^,^,^",options="header " , ] next we turn back to the set of 14 equations for the envelope functions @xmath131 , as given by the initial differential matrix ( 8) .
we find the envelope functions by substitution using an iteration procedure . in the first step we put @xmath57=0 and express twelve envelope functions by @xmath132 and @xmath133 , corresponding to the spin - up and spin - down @xmath39 conduction states . next the @xmath57 terms
are restored and the zero - order functions are put back to the complete equations .
the twelve first - order @xmath131 functions are expressed by @xmath132 and @xmath133 . in this approximation ,
linear in @xmath57 terms , the bia ( dresselhaus ) spin splitting is included . finally , we return again to the initial set and determine the twelve functions @xmath131 by @xmath132 and @xmath133 including the @xmath134 terms . the resulting equations for @xmath132 and @xmath133 represent the effective hamiltonian .
it is obtained in the form of a 2@xmath582 differential matrix since the bulk inversion asymmetry mixes the two spin states .
we are now in a position to calculate electron energies and the resulting conduction @xmath1-factor for given values of @xmath81 and @xmath9 ( still for the bulk ) . in the obtained formulas for energies there appear @xmath9 , @xmath78 and @xmath135 terms .
we checked that the @xmath9 and @xmath135 terms have negligible influence on the final energies .
theoretical spin @xmath1-factor of conducting electrons in bulk gaas _ versus _ the wave vector @xmath102 , or the corresponding well width @xmath101 ( see text ) , as calculated from the five - level model .
i - results for q = 0 ( first iteration step ) , i+ii - bulk inversion asymmetry included ( linear q terms , second iteration step ) , i+ii+iii - quadratic q terms included ( third iteration step ) .
the line marked ex shows `` exact '' results calculated numerically for the bulk by the evtuhov method . ] in fig .
3 we plot the calculated @xmath1-values as functions of @xmath101 ( or @xmath102 ) for the three consecutive approximations . to determine @xmath102 for a given width @xmath101 we use the same procedure as described before ( cf .
our three approximations for the @xmath1-value correspond to the three iteration steps .
curve i corresponds to @xmath57=0 , the corresponding formulas contain @xmath85 and @xmath136 type of terms .
curve i+ii includes the previous approximation plus the terms @xmath137 , @xmath138 , @xmath139 , accounting for the spin splitting due to bia .
curve i+ii+iii includes the two previous approximations plus the terms @xmath140 , @xmath141 , @xmath142 ( the last terms are very small and may be neglected ) . in the same fig .
3 we plot the @xmath1 value obtained from the `` exact '' numerical procedure based on matrix ( 21 ) . a discussion of the results shown in fig .
3 is in order .
first , curve i gives the results very similar to those marked by ex in fig .
although the calculation i includes the higher conduction levels , as compared to 3lm used for ex , it turns out that the results of 3lm and 5lm with @xmath57=0 are very similar if one adjusts @xmath97 and @xmath98 constants to have the same band - edge values of @xmath83 and @xmath84 .
second , sizable difference between the calculations i and i+ii may appear unexpected .
compared to i , the calculation i+ii includes the dresselhaus spin splitting due to bia , which is usually rather small . indeed , for relatively large @xmath101 ( small @xmath102 ) the corrections due to bia are small .
it is at small widths @xmath101 ( large @xmath102 ) that they become appreciable .
one should keep in mind that bia works directly on the spin splitting , i.e. on the @xmath1 value . more generally , the bia spin splitting has somewhat particular properties since , while it is _ affected _ by a magnetic field , its primary origin is _ not _ a magnetic field .
finally , it follows from fig .
3 that the curve i+ii+iii almost coincides with the `` exact '' result .
this means that the third step in the iteration procedure gives enough precision in the calculation of the electron @xmath1-value .
we use this information in the following treatment of the spin splitting in quantum wells .
we apply the above results to calculation of the spin @xmath1-factor for conduction electrons in qws grown along the * z * direction .
we use 5lm and consider both the growth direction and a magnetic field parallel to the [ 001 ] crystal axis .
matrix ( 8) must now be completed by the potential @xmath143 in all diagonal terms . the calculation is carried out by iterating solutions of eq .
( 8) , as described above for the bulk .
the difference is that now we keep the operator @xmath144 in the differential form since the envelope functions in qws are unknown and they must be determined by solving the eigenenergy problem . as a consequence ,
it is necessary in the iteration procedure to observe the order of @xmath145-dependent terms . in our treatment
we keep the free - electron and far - band contributions only in the conduction band . while a more general treatment is possible , it makes the final formulas very lengthy while the precision is not markedly higher . after a considerable manipulation the effective hamiltonian for the spin - up and spin - down electron states in the conduction band
is obtained in the form @xmath146\;\;,\ ] ] where @xmath147 @xmath148\frac{\partial}{\partial z}+ \right.\ ] ] @xmath149\right\}\;\;,\ ] ] @xmath150 @xmath151\frac{\partial}{\partial z } \right.+\ ] ] @xmath152\right\}\;\;.\ ] ] the effective mass is @xmath153 \;\;,\ ] ] and the @xmath1-factor is @xmath154 \;\;,\ ] ] where @xmath155 @xmath156 we use the notation @xmath157 , @xmath158 and @xmath159 . further @xmath160
@xmath161 @xmath162 @xmath163 @xmath164 @xmath165 @xmath166 @xmath167 @xmath168 @xmath169 we emphasize that the mass @xmath170 and the g - value @xmath171 defined in eqs .
( 25 ) and ( 26 ) do not represent the final cyclotron mass and g - value in a quantum well but only the first iterative approximations to these quantities , as obtained from matrix ( 8) by putting q = 0 . the nondiagonal component in eq .
( 22 ) , related to the bulk inversion asymmetry , is @xmath172 where @xmath173\;\;,\ ] ] @xmath174 in which @xmath175 @xmath176\right\}\;\;.\ ] ] the diagonal components in the hamiltonian ( 22 ) are composed of the terms proportional to @xmath85 and @xmath136 , resulting from the first iteration step ( marked i in the previous sections ) .
they contain also the terms proportional to @xmath140 and @xmath141 , resulting from the third iteration step ( marked iii ) .
the diagonal terms neither raise nor lower the harmonic oscillator functions . the nondiagonal terms @xmath177 and @xmath178 are proportional to @xmath57 and they come from the second iteration step ( marked ii ) .
they result from the bulk inversion asymmetry of the crystal .
the operator @xmath177 raises the harmonic oscillator function @xmath179 to @xmath180 , whereas @xmath181 lowers it from @xmath179 to @xmath182 . in consequence , solving the eigenenergy problem imposed by eq .
( 22 ) requires again the evtuhov procedure .
however , now we have to look for solutions in the form band structure of a rectangular gaas / ga@xmath5al@xmath6as quantum well within 5lm along the growth direction @xmath145 .
the calculated energy and wavefunction of the ground subband is indicated . ]
@xmath183 where @xmath184=1 , 2 , the functions @xmath185 are the same as for the bulk , but @xmath186 are as yet unknown envelope functions describing the motion along the growth direction @xmath145 .
when applying the functions ( 34 ) we will limit them to the minimal coupling scheme .
it can be seen from eq .
( 22 ) that the spin - up state @xmath187 described by @xmath188 interacts via the offdiagonal elements @xmath177 and @xmath181 with the spin - down states . in view of the above considerations it is clear that @xmath187 interacts via @xmath178 with @xmath189 and via @xmath177 with @xmath190 . after performing the operations on the harmonic oscillator functions we obtain the following eigenenergy problem in the minimal coupling scheme @xmath191
this set of coupled differential equations gives three energies , of which we are interested in @xmath192 .
a similar reasoning leads to the following set of differential equations containing the @xmath193 state @xmath194 here we are interested in @xmath195 . in the above matrices we use the following notation @xmath196 @xmath197\frac{\partial}{\partial z}+\right.\ ] ] @xmath198\right\}\ ] ] @xmath199 @xmath200\frac{\partial}{\partial z}+\right.\ ] ] @xmath201\right\}\ ] ] @xmath202 @xmath203 in which @xmath204 are again given by eq .
( 29 ) .
if one is interested in the lowest spin states @xmath205 , sets ( 35 ) and ( 36 ) do not contain the @xmath206 and @xmath207 components and the eigenenergy problems reduce to the following two sets of coupled differential equations . for the @xmath208 state @xmath209 while for the @xmath210 state @xmath211 in order to determine the cyclotron mass for the lowest levels we need to know in addition the energies of @xmath212 and @xmath213 states .
as follows from eq .
( 36 ) , for the @xmath213 state the corresponding differential set still contains two equations , while for the @xmath214 state we have in principle three coupled equations , see eq .
( 35 ) . however , we found that the addition of the third state @xmath215 ( see eq .
( 35 ) ) does not really change the energy of the @xmath214 state , so that two equations suffice to determine its energy .
once the electron is in a qw of a finite height , its total @xmath1-factor is given not only by its properties in the well but also by those in the barriers .
the situation for a rectangular gaas / ga@xmath5al@xmath6as qw is shown in fig .
4 . since the electron wave function penetrates into the barriers in which @xmath1 is different from those in the well , we should average over the two regions .
seemingly , one should simply use the band edge values of @xmath1 in gaas and ga@xmath5al@xmath6as and take an appropriate average .
in fact , this was done in ref .
3 . we argue , however , that this is incorrect .
the reason is that , as follows from fig .
4 , the electron energy is much lower than the barrier s band edge @xmath216 .
this effect is automatically included in our formalism if we use for both the well and the barriers the appropriate functions @xmath217 and @xmath218 , as given by eqs .
( 27 ) and ( 28 ) , respectively . when considering the two regions we have to change not only the appropriate values of energy gaps and spin - orbit energies but also the potential @xmath143 . at the flat bottom of the well shown in fig .
4 , there is @xmath143=0 . on the other hand , in the barrier regions
there is @xmath219 , where @xmath216 is the offset value for the conduction band between the two materials .
clearly , the energy @xmath13 is the same in the two regions .
only if @xmath216 were zero would we deal with `` straight '' values of @xmath1 in both materials .
it is instructive to consider a numerical example .
taking for ga@xmath5al@xmath6as the parameters given in table 1 we obtain the band edge value of @xmath84 = 0.5378 , in agreement with the experimental findings [ 13 ] . on the other hand , when we compute the value of @xmath1 in the barrier using the same parameters but accounting for @xmath216 = 0.264mev and taking to a good approximation @xmath220 0 , we obtain @xmath1 = 0.1638 , which is a considerably lower value . in fact , in our formalism we do not employ the averaging procedure . when calculating the total electron @xmath1 value we use in the expressions for @xmath221 , @xmath222 and @xmath223 , as given by eqs ( 37 ) , ( 38 ) and ( 29 ) , the actual functions @xmath217 and @xmath218 for a given region .
this includes @xmath143 and the local values of band parameters .
thus the actual values of @xmath171 , @xmath170 and @xmath223 enter both the differential equations defined by eq .
( 22 ) , as well as the boundary conditions at the interfaces .
the problem of the boundary conditions was intensely discussed in the early days of heterostructure investigations , see [ 4 ] .
it will suffice to say here that our eqs .
( 37 ) and ( 38 ) are written in the hermitian form , so that the boundary conditions are obtained in the form of continuity of the wave function and of electric current across the interfaces .
the eigenenergy calculation is done for each spin separately so that , as follows from eq .
( 22 ) , also the boundary conditions depend on the spin .
since the effective equations for the energies contain only @xmath224 terms , the boundary conditions are not more complicated than those for a @xmath145-dependent effective mass .
once @xmath225 and @xmath226 are computed , the @xmath1-value is determined using the definition ( 15 ) .
all the above remarks apply also to the computation of the cyclotron mass .
the mass value in barriers also contains the offset @xmath216 . in the above numerical example for ga@xmath5al@xmath6as , instead of the band - edge value @xmath83 = 0.0875@xmath99 we deal with the value @xmath227 = 0.0748@xmath99 in the barrier .
we emphasize that this feature is of importance even in the simplest calculation of the energy and of the effective mass in a qw made of two materials .
the above problem is recognized in the work of bastard et al [ 19 , 20 ] as well as in our earlier papers [ 21 , 22 ] .
figure 5 shows our calculation of the electron spin @xmath1-factor in gaas / ga@xmath5al@xmath6as rectangular qws together with available experimental data .
this is the main result of our work .
it can be seen that the complete five - level * p@xmath0p * model gives an excellent description of the @xmath1-value as a function of the well width between 3 nm and 12 nm . in particular
, the theory reproduces the experimental value of @xmath1 = 0 for the well width of 6 nm , as observed by le jeune et al [ 23 ] . in the discrepancy of the experimental points around d @xmath228 4.5 nm ,
priority should be given to ref .
[ 23 ] ( the full circle ) since the raman data of ref .
[ 24 ] ( the empty circle ) do not determine the conduction @xmath1-value directly .
our calculation is carried out for the growth direction @xmath229 $ ] with an external magnetic field parallel to this direction . in order to reach the good description of experiments we had to use the full five - level model of the band structure , as discussed above .
the theory agrees perfectly well with the experimental @xmath1-value even for @xmath230 30 @xmath231
. one could find it surprising that the * p@xmath0p * theory works so well on the scale of few interatomic distances in gaas , but this result is in agreement with a more general experience that the * p@xmath0p * theory works better than it should .
spin @xmath1-factor of conduction electrons in rectangular gaas / ga@xmath5al@xmath6as quantum wells _ versus _ well width @xmath101 .
the solid line - theory based on 5lm , the dashed line - theory based on 3lm , full circles - experimental data of le jeune et al [ 23 ] , empty circles - experimental data of sapega et al [ 24 ] . ]
it can be seen that the total @xmath1-value for narrow wells is smaller than the value for pure gaas , as shown in fig .
this is due to the above mentioned effect of the offset @xmath216 , which lowers the effective @xmath1 in the barriers , as compared to the band - edge @xmath1-value in ga@xmath5al@xmath6as . the dashed line in fig .
5 shows the calculation of @xmath1-factor , in which we used the three - level * p@xmath0p * model . this model does not involve the @xmath57 matrix element , so that it includes neither bia splitting nor the @xmath134 terms .
it can be seen that 3lm does not properly describe the experimental data in gaas / ga@xmath5al@xmath6as quantum wells .
this result agrees with the general conclusion that the three - level model does not describe correctly the conduction band in gaas - type materials , see [ 1 , 2 ] .
the values of @xmath1 shown in fig . 5 were computed for the magnetic field intensity @xmath81 = 2 t. it is natural that in a theory for the nonparabolic energy band , when the energies are nonlinear functions of a magnetic field , the @xmath1-value may depend somewhat on @xmath81 .
we checked this dependence and found that diminishing the field until @xmath232 0.25 t the @xmath1-value varies only weakly with @xmath81 .
however , as @xmath81 tends to zero the @xmath1-value ceases to be a useful quantity because of the spin splitting due to bia .
it is then more practical to use the energy splitting for the two spins . as to the previous description of @xmath1-value in qw based on 3lm [ 3 ]
, it is incorrect on two accounts . first , as already mentioned in the discussion of fig .
2 , it predicts a much too strong increase of @xmath1 in gaas with decreasing well width @xmath101 .
second , when averaging the spin @xmath1-factors in gaas well and ga@xmath5al@xmath6as barriers , it takes the band - edge values in both materials , neglecting the influence of the band offset on the @xmath1-value in barriers .
both above shortcomings lead to an increase of the total @xmath1 . we conclude that the agreement of the theory [ 3 ] with the experiment on gaas / ga@xmath5al@xmath6as qw , as quoted by le jeune et al [ 23 ] , is fortuitous .
cyclotron mass of conduction electrons in rectangular gaas / ga@xmath7al@xmath8as quantum wells _ versus _ well width @xmath101 .
the solid line - theory based on 5lm , the dashed line - theory , based on 3lm , experimental data are after huant et al [ 25 ] for samples with slightly different chemical composition .
full circles x=0.26 , full square x=0.25 , full triangle x=0.23 . ] in fig .
6 we plot the cyclotron mass of electrons in gaas / ga@xmath7al@xmath8as rectangular qws as a function of the well width according to the three - level and five - level models .
it can be seen that 3lm gives distinctly lower masses , although we adjust the far - band contribution @xmath97 to get the same band - edge value of @xmath83 = 0.0665 @xmath99 .
the five - level model gives slightly different masses for spin - up and spin - down cyclotron transitions .
figure 6 shows the comparison of our theory with the cyclotron resonance data of huant et al [ 25 ] obtained on gaas / ga@xmath3al@xmath4as qws .
for the gaas / ga@xmath7al@xmath8as alloy we use the band parameters indicated in table 1 . for the band edge in gaas
we take the perfectly acceptable value of @xmath83 = 0.0665 @xmath99 , accounting for the fact that the data of ref .
[ 25 ] were taken at the temperatures 20 - 30 k. our theory describes the data very well with the exception of the point at @xmath230 20 nm , but this point does not follow well the overall mass behavior .
the calculation shown in fig .
6 was performed for @xmath81 = 10 t , which corresponds to the experimental conditions of ref .
[ 25 ] . comparing the data with the theory
we conclude that the three - level model can not correctly describe the experimental data . the previous theoretical treatment of the effective masses in gaas / ga@xmath3al@xmath4as qws by ekenberg [ 4 ] , which was carried out for @xmath81 = 0 , makes a distinction between the masses parallel and perpendicular to the growth direction .
we do not introduce the parallel mass since it does not seem to have a clear physical meaning .
also , in nonparabolic energy bands one can not separate the electron motion in different directions since an increase of the mass due to motion in one direction will affect the motion in other directions .
the theory of ref .
[ 4 ] is based on an expansion of the energy in powers of momentum [ 26 ] .
the coefficients in such an expansion contain the band - edge energies in their denominators and the interband matrix elements of momentum in their numerators . in ref .
[ 4 ] the potential @xmath143 of the well is added later , so that this procedure can be qualified as semiclassical . on the other hand , in our treatment
we introduce the potential @xmath143 and the energy @xmath100 from the beginning and use the iteration procedure in solving the resulting 14 equations by substitution .
this procedure is exact for the matrix element @xmath57 = 0 and includes @xmath57 to second order .
the resulting effective equations have the coefficients which not only contain the band - edge energies , but also involve in their denominators @xmath143 and @xmath100 , cf .
( 27 ) , ( 28 ) , ( 29 ) .
we showed above that these quantities are not negligible in gaas - type materials and they would be even more important in narrow - gap materials .
the work [ 4 ] does not include the bulk inversion asymmetry effects , which are not very important for the cyclotron mass but become important for the spin @xmath1-value , see our fig .
3 . our theoretical masses for gaas / ga@xmath233al@xmath234as qw ( not shown ) are somewhat higher then those presented in fig . 4 of ref .
it is not clear whether the differences result from the approximations made in ref .
[ 4 ] or from somewhat different band parameters used in the two calculations .
the approach presented above can be used equally well for quantum wells described by an arbitrary potential @xmath143 .
this amounts to solving differential equations ( 22 ) , ( 37 ) , ( 38 ) , in which @xmath143 explicitly appears .
however , if a well ( and a corresponding potential ) does not possess the inversion symmetry , there appears a spin splitting due to the bychkov - rashba mechanism [ 6 ] .
technically , this appears as a result of noncommutativity of @xmath235 with @xmath143 when one solves the * p@xmath0p * equations by substitution .
the bychkov - rashba splitting has been treated in many papers , see the review [ 27 ] .
the present work confirms our previous description of the spin @xmath1-factor in parabolic gaas / ga@xmath3al@xmath4as quantum wells [ 28 , 29 ] .
we conclude that the five - level * p@xmath0p * model adequately describes this important system .
we describe the spin and cyclotron energies of electrons in gaas / ga@xmath3al@xmath4as quantum wells in the presence of an external magnetic field parallel to the growth direction [ 001 ] .
our approach is based on the five - level * p@xmath0p * model of band structure for gaas - type materials .
inadequacy of the previous theory used in the literature for the spin @xmath1-factor in heterostructures is indicated .
we solve 14 coupled differential equations resulting from the * p@xmath0p * formulation by three iteration steps .
the sufficiency of this procedure is tested on bulk gaas energies for different values of @xmath10 corresponding to real quantum wells .
influence of the bulk inversion asymmetry present in iii - v compounds on the electron @xmath1-value is emphasized .
our theory gives an excellent description of the existing experimental data on the spin @xmath1-factor and the cyclotron mass of electrons in gaas / ga@xmath3al@xmath4as rectangular quantum wells for different well widths .
|
a five - level * p@xmath0p * model of the band structure for gaas - type semiconductors is used to describe the spin @xmath1-factor and the cyclotron mass @xmath2 of conduction electrons in gaas / ga@xmath3al@xmath4as quantum wells in an external magnetic field parallel to the growth direction .
it is demonstrated that the previous theory of the @xmath1-factor in heterostructures is inadequate .
our approach is based on an iteration procedure of solving 14 coupled differential * p@xmath0p * equations . the applicability of the iteration procedure is verified . the final eigenenergy problem for the conduction subbands
is reduced to two differential equations for the spin - up and spin - down states of consecutive landau levels .
it is shown that the bulk inversion asymmetry of iii - v compounds is of importance for the spin @xmath1-factor .
our theory with no adjustable parameters gives an excellent description of experimental data on the electron spin @xmath1-factor in gaas / ga@xmath5al@xmath6as rectangular quantum wells for different well widths between 3 and 12 nm .
the same theory describes very well experimental cyclotron masses in gaas / ga@xmath7al@xmath8as quantum wells for the well widths between 6 and 37 nm .
|
lattice - gauge calculations yield a non - vanishing value of the quark condensate @xmath3 in the qcd vacuum @xcite , indicating that chiral symmetry is broken .
this general feature of the vacuum remains present even for massless quarks because the symmetry is then broken spontaneously . on the other hand
, chiral symmetry is expected to become restored at sufficiently high values of the net - baryon density @xmath4 or / and the temperature @xmath5 .
the character of this phase change is not yet well understood but it has significant implications in areas such as cosmology and astrophysics and it is a focal point for current experimental and theoretical research in nuclear physics . nuclear collision experiments carried out with the relativistic heavy ion collider ( rhic ) at the brookhaven national laboratory and with the large hadron collider at cern explore systems having relatively small net - baryon densities @xmath4 and the associated chemical potentials @xmath6 are negligible .
lattice calculations can readily be carried out at vanishing @xmath6 and they indicate that a cross - over transformation from the chirally broken phase to the restored phase occurs as the temperature is increased from below to above the cross - over temperature @xmath7 @xcite . the other extreme region of the qcd phase diagram , namely low temperatures and high chemical potentials , can not be addressed by current lattice - qcd methods , due to the fermion sign problem , and studies of this phase region must therefore rely on less fundamental models .
most investigations suggest that there is a first - order phase transition which , for @xmath8 , sets in at baryon densities several times that of the nuclear saturation density , @xmath9 .
the properties of strongly interacting matter in this phase region are important for our understanding of compact stars .
if indeed such a first - order phase transition exists at @xmath10 , then , as the temperature is raised , one would expect it to remain present but gradually weaken and eventually terminate at a critical point @xmath11 .
the existence and location of such a critical point is a subject of intense theoretical investigation with a variety of models , including in particular effective - field models , such as the linear @xmath1 model ( lsm ) , and effective quark models , such as the nambu jona - lasinio ( njl ) , at different levels of sophistication considering up to three quark flavors and possibly including the polyakov loop to account for confinement @xcite .
experimentally the corresponding region of density and excitation may be produced in current nuclear collisions at the low - energy end of rhic and in the future with fair at gsi and nica at jinr which are being constructed with such investigations in mind . in the present work ,
we concentrate on the high-@xmath6 and low-@xmath5 part of the phase diagram with the aim of exploring the expected chiral phase transition which has significant implications for the possible existence of quark stars @xcite .
it should be noted that chiral symmetry may be restored already during the early post - bounce accretion stage of a core - collapse supernova event and the associated neutrino burst might then provide a spectacular signature for the presence of quark matter inside compact stars @xcite .
however , as pointed out in refs .
@xcite , the possibilities depend on the dynamics of the phase conversion and especially on the time scales involved .
when the phase diagram of bulk matter exhibits a first - order phase transition , the two phases may coexist in mutual thermodynamic equilibrium and , consequently , when brought into physical contact a mechanically stable interface will develop between them .
the associated interface tension @xmath12 ( which we shall often refer to simply as the surface tension of quark matter ) depends on the temperature @xmath5 ; it has its largest magnitude at @xmath10 and approaches zero as @xmath5 is increased to @xmath13 .
this quantity plays a key role in the phase conversion process and it is related to various characteristic quantities such as the nucleation rate , the critical bubble radius , and the favored scale of the blobs generated by the spinodal instabilities @xcite .
( as we shall see , the surface tension is essentially proportional to the effective interaction range , which determines the width of the surface region , and the spatial size of the most rapidly amplified density irregularity is also proportional to this quantity . ) unfortunately , despite its central importance , the surface tension of quark matter is rather poorly known .
estimates in the literature fall within a wide range , typically @xmath14 @xcite and values of @xmath15 have been considered for studying the effect of quark matter nucleation on the evolution of proto - neutron stars @xcite .
but the authors in ref .
@xcite , taking into account the effects from charge screening and structured mixed phases , estimate @xmath16 , without excluding smaller values , and an ever higher value , @xmath17 , was found by alford _
et al . _
@xcite on the basis of dimensional analysis of the minimal interface between a color - flavor locked phase and nuclear matter .
the surface tension for two - flavor quark matter was evaluated within the framework of the lsm by palhares and fraga @xcite . in that work
the authors considered the one - loop effective potential and then fitted its relevant part , which included both the chirally symmetric and broken state , by a quartic polynomial .
the surface tension was evaluated using the thin - wall approximation for bubble nucleation and the estimated values cover the @xmath18 range , depending on the inclusion of vacuum and/or thermal corrections . in principle , this range makes nucleation of quark matter possible during the early post - bounce stage of core - collapse supernovae and it is thus a rather important result .
it is also worth noting that a small surface tension would facilitate various structures in compact stars , including mixed phases @xcite .
the present work is devoted to the evaluation of the surface tension for quark matter using both the lsm ( with two flavors ) and the njl model ( with two and three flavors ) following the procedure employed in ref .
@xcite . here , the lsm is mainly included to check the consistency of our procedure by comparing our present results with those obtained by the thin - wall approximation of ref .
@xcite ( we find the agreement to be very good ) . the njl model is considered with two and three flavors because the latter , which contains strangeness , is one of the most popular effective quark models used in studies related to compact stars . as explained below , the method described in ref .
@xcite makes it possible to express the surface tension for any subcritical temperature in terms of the free energy density for uniform matter in the unstable density range . because the models employed readily provide the equation of state ( eos ) for the full density range , they are well suited for our purpose and we may directly employ the method without any further approximations . in practice ,
the procedure is rather simple to implement and it provides an estimate for the surface tension that is consistent with the eos implied by the model employed , with its specific approximations and parametrizations .
the paper is organized as follows . in sect .
[ method ] we review the method for extracting the surface tension from the equation of state . in sec . [ models ]
we then present the two - flavor versions of the two models considered and discuss how to extract the surface tension .
section [ njl3 ] is devoted to the treatment for the more realistic @xmath19 version of the njl model and our numerical results are presented in sec .
[ results ] , both for cold matter and for temperatures up to the critical value . the conclusions and final remarks are presented in sec .
[ conclude ] .
we assume here that the material at hand , strongly interacting matter , may appear in two different phases under the same thermodynamic conditions of temperature @xmath5 , chemical potential @xmath6 , and pressure @xmath20 .
these two coexisting phases have different values of other relevant quantities , such as the energy density @xmath21 , the ( net baryon ) density @xmath4 , and the entropy density @xmath22 . under such circumstances
, the two phases will develop a mechanically stable interface if placed in physical contact and it is the purpose of the present study to evaluate the associated interface tension , @xmath12 .
the two - phase feature appears for all temperatures below the critical value , @xmath13 .
thus , for any subcritical temperature , @xmath23 , hadronic matter at the ( net - baryon ) density @xmath24 has the same chemical potential and pressure as quark matter at the ( larger ) density @xmath25 . as @xmath5
is increased from zero to @xmath13 , the coexistence phase points @xmath26 and @xmath27 trace out the lower and higher branches of the phase coexistence boundary , respectively , gradually approaching each other and finally coinciding for @xmath28 . any @xmath29 phase point outside of this boundary corresponds to thermodynamically stable uniform matter , whereas uniform matter prepared with a density and temperature corresponding to a phase point inside the phase coexistence boundary is thermodynamically unstable and prefers to separate into two coexisting thermodynamically stable phases separated by a mechanically stable interface .
because such a two - phase configuration is in global thermodynamic equilibrium , the local values of @xmath5 , @xmath6 , and @xmath20 remain unchanged as one moves from the interior of one phase through the interface region and into the interior of the partner phase , as the local density @xmath4 increases steadily from the lower coexistence value @xmath30 to the corresponding higher coexistence value @xmath31 .
it is convenient to work in the canonical framework in which where the control parameters are temperature and density .
the basic thermodynamic function is thus @xmath32 , the free energy density as a function of the ( net baryon ) density @xmath4 for the specified temperature @xmath5 .
the chemical potential can then be recovered as @xmath33 , and the entropy density as @xmath34 , so the energy density is @xmath35 , while the pressure is @xmath36 . for single - phase systems
@xmath32 is convex , _
i.e. _ its second derivative @xmath37 is positive , while the appearance of a concavity in @xmath32 signals the occurrence of phase coexistence , at that temperature .
this is easily understood because when @xmath32 has a local concave anomaly then there exist a pair of densities , @xmath30 and @xmath31 , for which the tangents to @xmath32 are common .
therefore @xmath32 has the same slope at those two densities , so the corresponding chemical potentials are equal , @xmath38 .
furthermore , because a linear extrapolation of @xmath32 leads from one of the touching points to the other , also the two pressures are equal , @xmath39
. so uniform matter at the density @xmath30 has the same temperature , chemical potential , and pressure as uniform matter at the density @xmath31 .
the common tangent between the two coexistence points corresponds to the familiar maxwell construction and shall here be denoted as @xmath40 .
obviously , @xmath32 and @xmath40 coincide at the two coexistence densities and , furthermore , @xmath32 exceeds @xmath40 for intermediate densities
. therefore we have @xmath41 . for a given ( subcritical ) temperature @xmath5
, we now consider a configuration in which the two coexisting bulk phases are placed in physical contact along a planar interface .
the associated equilibrium profile density is denoted by @xmath42 where @xmath43 denotes the location in the direction normal to the interface . in the diffuse interface region ,
the corresponding local free energy density , @xmath44 , differs from what it would be for the corresponding maxwell system , _
i.e. _ a mathematical mix of the two coexisting bulk phases with the mixing ratio adjusted to yield an average density equal to the local value @xmath45
. this local deficit amounts to @xmath46 where @xmath47 is either one of the two coexistence densities .
the function @xmath48 is smooth and it tends quickly to zero away from the interface where @xmath42 rapidly approaches @xmath47 and @xmath44 rapidly approaches @xmath49 .
the interface tension @xmath12 is the total deficit in free energy per unit area of planar interface , @xmath50 as discussed in ref .
@xcite , when a gradient term used to take account of finite - range effects , the tension associated with the interface between the two phases can be expressed without explicit knowledge about the profile functions but exclusively in terms of the equation of state for uniform ( albeit unstable ) matter , @xmath51^{1/2 } \frac { d \rho}{\rho_g}\ , \ ] ] where @xmath52 is a characteristic value of the density and @xmath53 is a characteristic value of the energy density , while the parameter @xmath54 is an effective interaction range related to the strength of the gradient term , @xmath55 .
we choose the characteristic phase point to be in the middle of the coexistence region , @xmath56 and @xmath57/2 $ ] , where @xmath58 is energy density at @xmath59 , while @xmath60 is energy density at the critical point @xmath61 .
the length @xmath54 as a somewhat adjustable parameter governing the width of the interface region and the magnitude of the tension @xcite . for the lsm it is natural to expect that @xmath62 which , also , is approximately the value found in an application of the thomas - fermi approximation to the njl model @xcite
. therefore , we shall adopt the value @xmath63 throughout the present work . while there is some arbitrariness in fixing these quantities ,
it is reassuring that the resulting surface tension is in excellent agreement with the value obtained in ref .
@xcite . with these parameters fixed
, the interface tension can be calculated once the free energy density @xmath32 is known for uniform matter in the unstable phase region , @xmath64 . while this is straightforward in a canonical formulation , where each @xmath29 characterizes only one manifestation of the system , even inside the unstable phase region , the task is more complicated in the commonly used grand canonical formulation because a given @xmath65 phase point characterizes three different manifestations of the system , one stable , one metastable , and one unstable .
the metastable solutions are located near the coexistence densities , while the unstable solutions are located in the intermediate spinodal region where uniform matter is mechanically unstable so that even infinitesimal irregularities may be exponentially amplified . by contrast
, only irregularities of a sufficient amplitude are amplified in the metastable regions , leading towards either nucleation ( near the lower coexistence density @xmath30 ) or bubble formation ( near the higher coexistence density @xmath31 ) .
in this section , we review the mean - field results for the thermodynamic potential for the two effective models when only two quark flavors are included .
these results have been widely discussed in the literature and here we follow ref .
@xcite ( see ref .
@xcite for results beyond the mean - field approximation ) .
the two models are similar in the sense that they do not have confinement and they incorporate spontaneous chiral symmetry breaking , which happens at the classical level in the lsm but only via quantum corrections in the njl model .
the fermionic fields representing the quarks are the only degrees of freedom in the njl model at the tree level , while the lsm also contains scalar ( @xmath1 ) and pseudoscalar ( @xmath66 ) meson fields .
in standard notation , the lagrangian density of the lsm with quarks reads @xmath67\psi\ ; , \label{lsm1}\ ] ] where @xmath68 is the flavor isodoublet spinor representing the quarks ( @xmath69 and @xmath70 ) , and @xmath71 is the classical potential energy density . in the chiral limit ( obtained for @xmath72 ) the chiral symmetry , @xmath73 ,
is spontaneously broken at the classical level , and the pion is the associated massless goldstone boson . for @xmath74 , the chiral symmetry is explicitly broken by the last term in @xmath75 which gives the pion a finite mass at vanishing @xmath5 and @xmath6 .
the parameters are usually chosen so that chiral symmetry is spontaneously broken in the vacuum and the expectation values of the meson fields are @xmath76 and @xmath77 where @xmath78 is the pion decay constant .
following ref .
@xcite , we fix the parameters as follows : @xmath79 , @xmath80 , and @xmath81 . using the standard relations , @xmath82 , @xmath83 , and @xmath84 , we obtain the meson masses , @xmath85 and @xmath86 .
the coupling constant @xmath87 is usually fixed so that the effective quark mass in vacuum , @xmath88 , be about one third of the nucleon mass , which gives @xmath89 .
we note that the same parameter set was also used to evaluate the surface tension , @xmath12 , in ref .
@xcite . to the one - loop level ,
the grand canonical potential is obtained by integrating the action over the fermionic fields @xcite , @xmath90 -t\ln \left [ 1- n^-\right ] \right\}\ , \ ] ] where @xmath91 , @xmath92 , @xmath93 , and @xmath94\}^{-1}$ ] represent the particle / antiparticle thermal occupancies with @xmath95 , where @xmath96 . for for given values of @xmath5 and @xmath6 ,
the equilibrium values of the meson fields are obtained by minimizing @xmath97 with respect to those , yielding the most likely values @xmath98 and @xmath99 .
the latter one vanishes in the mean - field approximation , so the associated constituent quark mass is given as @xmath100 .
the minimum value of the grand potential represents minus the equilibrium pressure , @xmath101 , so the net quark density is given by @xmath102 , where @xmath103 . the entropy density given by @xmath104 , while the energy density , @xmath21 , can then be obtained by means of the standard thermodynamic relation @xmath105 and the free energy density is @xmath106 . in the neighborhood of the phase coexistence line in the @xmath65 plane ,
the grand potential has three extrema representing stable , metastable , and spinodally unstable matter .
as emphasized above , the extraction of the surface tension by the geometric approach requires the consideration of all three extrema .
in contrast to the njl model , the vacuum term represented by the first term in the integrand of eq .
( [ omegalsm ] ) is not essential for the spontaneous breaking of chiral symmetry . in the lsm
this already happens at the classical level and the symmetry restoration is driven mainly by the terms containing @xmath107 .
therefore , we neglect the vacuum term in the present lsm application , where the aim is to compare our estimates with the zero - temperature interface tension obtained in ref .
@xcite , @xmath108 , where the relevant part of the same thermodynamical potential was fitted with a quartic polynomial . in our approach
such a fitting procedure is not necessary because the thermodynamic potential is evaluated for all values of @xmath6 and @xmath5 .
this will lead to somewhat different numerical values for the surface tension @xmath12 .
it was shown in ref .
@xcite that the inclusion of vacuum terms at @xmath10 increases the surface tension value from @xmath109 to about @xmath110 . in practice , further refinements including vacuum and in - medium two - loop corrections are possible by following the same technical steps that were employed in the evaluation of the thermodynamical potential for the yukawa theory @xcite .
it is now straightforward to determine the phase - coexistence line in the @xmath65 plane which forms the starting point for determining all quantities related to @xmath12 .
it starts at @xmath111 and terminates at the critical point @xmath112 , which agrees with ref .
we now have all the ingredients needed for determining the coexistence densities @xmath24 and @xmath25 as well as the characteristic values @xmath52 and @xmath53 appearing in eq .
( [ gamma ] ) . the difference @xmath113 can be readily determined numerically by considering the stable ( global ) minimum , the metastable ( local ) minimum , and the unstable ( local ) maximum appearing in the thermodynamical potential , as will be explicitely shown in section [ results ] .
although our method for extracting the surface tension does not require the profiles @xmath114 and @xmath45 , these functions do provide interesting additional information about the interface . to obtain the profile functions within the lsm
, it suffices to consider the grand canonical potential in the @xmath1 direction only , _
i.e. _ taking @xmath115 .
at @xmath10 it can be expressed in terms of the fermi momentum @xmath116 ( given by @xmath117 ) , @xmath118 \right\}\ .\ ] ] we now employ the local density approximation , so the local fermi momentum , @xmath119 , is related to the local density , @xmath120 , by @xmath121 .
furthermore , the local scalar density @xmath122 is given by @xmath123\ , \ ] ] where we have used that the local fermi energy @xmath124 equals the ( constant ) chemical potential @xmath125 .
we note that @xmath126_{\sigma=\sigma({\boldsymbol{r } } ) } \ = \
u^\prime(\sigma({\boldsymbol{r } } ) ) + g \rho_s({\boldsymbol{r}})\ . \label{bubble2}\ ] ] then the stationary euler - lagrange equation with @xmath115 and @xmath127 replaced by @xmath128 , provides an equation for the local value of the order parameter , @xmath129 , @xmath130\ = \ 0\ . \label{motion}\ ] ] for semi - infinite geometry , the profile of the order parameter , @xmath114 , can then be obtained by solving the corresponding euler - lagrange equation , @xmath131_{\sigma=\sigma(z)}\ = \ 0\ , \label{bubble1}\ ] ] with the boundary conditions that the order parameter approach the zero - temperature coexistence values far from the surface , @xmath132 once @xmath114 is known , so is the mass @xmath133 , and we can then obtain the local fermi momentum @xmath134 and , consequently , the local net baryon density @xmath45 .
one may then define the associated interface location function @xcite , @xmath135 which allows us to obtain additional information , such as the mean interface location , @xmath136 , the associated interface width @xmath137 , where @xmath138 , as well as a measure of the profile skewness which is given by the dimensionless parameter @xmath139 @xcite .
the calculated results for @xmath114 , @xmath45 , and @xmath140 are shown in fig .
the origin is conveniently located at @xmath141 , the width is @xmath142 ( where @xmath143 ) and the skewness is @xmath144 . as the temperature
is increased , the profiles widen progressively and grow more symmetric , as also found in ref .
@xcite .
we now consider the standard version of the two - flavor njl model @xcite .
its lagrangian density is based on a chirally symmetric four - fermion interaction , @xmath145 , \label{njl2}\ ] ] where @xmath68 is to be interpreted as in the lsm .
furthermore , it is assumed that @xmath146 so the mass matrix is given by @xmath147 . in the mean - field approximation
, the grand canonical potential reads @xcite @xmath148-t\ln\left[1-n^-\right]\right\}\ , \label{omeganjl}\ ] ] with the same definitions as used in the lsm . for each value of @xmath5 and @xmath6
the dynamical mass is of the form @xmath149 because of the assumption of isospin symmetry ( @xmath150 ) and chemical equilibrium ( @xmath151 ) ; alternative scenarios may also be considered @xcite .
the single dynamical mass @xmath152 is then obtained by minimizing @xmath153 with respect to @xmath152 , leading to the well known gap equation , @xmath154\ .
\label{gap}\ ] ] although the thermodynamic potentials for lsm ( eq . ( [ omegalsm ] ) ) and the njl model ( eq . [ omeganjl ] ) )
have the same structure as far as the loop contribution is concerned , some important differences between the two models exist .
first , we note that within the njl the quark mass acquires its constituent value only when quantum corrections ( loop terms ) are computed . therefore , contrary to the lsm , the divergent term represented by the second term on the right - hand side of eq .
( [ omeganjl ] ) plays a central role regarding the ( dynamical ) chiral symmetry breaking .
another difference between the lsm and the njl model , in 3 + 1 dimensions , is that the latter is not renormalizable since the coupling @xmath155 carries dimensions ( @xmath156 ) .
this means that potential divergencies can not be systematically eliminated by a redefinition of the original parameters . by considering it as an effective model , one gives up the very high energies and evaluates all the integrals up to an ultraviolet ( non - covariant ) cutoff @xmath157 , as the notation in eqs .
( [ omeganjl ] ) and ( [ gap ] ) implies .
then @xmath157 is treated as a `` parameter '' which will be fixed , together with @xmath155 and @xmath158 , so as to yield the values of physical observables such as @xmath159 , @xmath160 that reproduce the phenomenological value of @xmath161 .
for example , in ref .
@xcite the authors reproduce @xmath162 and @xmath85 using @xmath163 and @xmath164 with @xmath165
. these parameter values , which we label `` set i '' , predict a first - order phase transition starting at @xmath166 and ending at the critical point @xmath167 , while the constituent quark mass in vacuum is @xmath168 . in their study of the chiral phase transition in the presence of spinodal decomposition
the authors in ref .
@xcite use @xmath169 and @xmath170 with @xmath171 in order to reproduce @xmath172 and @xmath173 obtaining @xmath174
. we shall also consider these parameter values , which we label `` set ii '' , in order to estimate the influence of different parametrizations in the estimation of @xmath12 .
with parameter set ii the first - order transition line starts at @xmath175 and ends at @xmath176 .
in general , a larger value of @xmath177 enlarges the coexistence region . as in the lsm case
, the quantities @xmath30 , @xmath31 , @xmath52 , @xmath53 , and @xmath113 entering the expression ( [ gamma ] ) for the surface tension can be obtained from the equation of state . as already emphasized ,
the numerical value for the length scale @xmath54 is chosen to be @xmath178 ( which is about the value found in a thomas - fermi application to the njl model @xcite ) .
the remaining two numerical inputs , @xmath52 and @xmath179 , are automatically fixed once the eos has been determined .
in stellar modeling , the structure of the star depends on the assumed eos built with appropriate models while the true ground state of matter remains a source of speculation .
it has been argued @xcite that _ strange quark matter _ ( sqm ) is the true ground state of all matter and this hypothesis is known as the bodmer - witten conjecture .
hence , the interior of neutron stars should be composed predominantly of @xmath180 quarks ( plus leptons if one wants to ensure charge neutrality which is not the case in the present work ) .
strangeness is implemented in the @xmath19 version of the njl model which is given by @xmath181
-k \left\{{\rm det}_f[\bar\psi(1+\gamma_5)\psi ] + { \rm det}_f[\bar\psi(1-\gamma_5)\psi]\right\}\ , \label{njl3}\ ] ] where @xmath182 denotes a quark field with three flavors ( and three colors ) , and @xmath183 is the corresponding mass matrix . here
we assume @xmath184 indicating that isospin symmetry is observed while the @xmath19 flavor symmetry is explicitely broken .
the eight gell - mann matrices are represented by @xmath185 ( @xmath186 ) and @xmath187 .
more details concering this version of the njl model can be found in ref .
@xcite . in the mean - field approximation
the thermodynamical potential is given by @xmath188 - 4k\phi_u\phi_d\phi_s\ .\ ] ] the term @xmath189 , which represents the contribution of a gas of quasiparticles with mass @xmath190 , is given by @xmath191-t\ln\left[1-n_q^-\right]\right\}\ , \ ] ] where @xmath192 and @xmath193\}^{-1}$ ] represent the particle / antiparticle distribution function .
for the quark condensates , @xmath194 , one has @xmath195\ .
\label{gap3}\ ] ] finally , the gap equation is @xmath196 which contains a non - flavor mixing term proportional to @xmath155 as well as a flavor mixing term proportional to @xmath197 . in our numerical analysis
we adopt the parameter values of ref .
@xcite which are @xmath198 , @xmath199 , @xmath200 , @xmath201 , and @xmath202 .
then , at @xmath10 and @xmath203 , one reproduces @xmath172 , @xmath204 , @xmath205 , and @xmath206 . for the quark condensates one obtains @xmath207 , and @xmath208 .
the constituent quark masses are then given by @xmath209 and @xmath210 .
the pressure , @xmath20 , and energy density , @xmath21 , follow from the usual expressions , @xmath211 here , for simplicity , we take the chemical equilibrium condition , @xmath212 which yields @xmath213 also at finite @xmath5 and/or @xmath6 .
of course , for a realistic description of neutron - star matter , charge and strangeness neutrality need to be taken into account , which is technically straightforward .
the phase diagram for this three - flavor quark model model in the @xmath6@xmath5 and @xmath4@xmath5 planes can be found in ref .
@xcite ; the first - order transition line starts at ( @xmath10 , @xmath214 ) and ends at the critical point ( @xmath215 , @xmath216 ) .
as already discussed , we need the eos inside the phase coexistence region which can be obtained by examining how the effective masses behave in this domain .
this behavior is shown in fig .
[ fig3 ] for @xmath10 ; these results go beyond those of ref .
@xcite by also considering the strange quark mass .
to understand this figure , let us recall that , in most situations , one is generally interested only in those solutions of the gap equation that correspond to global ( stable ) minima of the thermodynamical potential .
however , when a first - order phase transition is present , there are two different such solutions for the same thermodynamic conditions of temperature , chemical potential , and pressure ( corresponding to the solid dots on fig . [ fig3 ] ) . as the net baryon density
( which serves as a convenient order parameter ) is increased from its lower coexistence value @xmath30 to its higher coexistence value @xmath31 , the thermodynamically favored state is a maxwell mixture of the two coexisting phases and the overall average of the energy per net baryon or the effective mass , for example , evolve monotonically along the socalled maxwell line , as the composition of the mixture changes from being entirely one phase to being entirely the other . in the region between the dotted lines
the gap equation has three solutions , leading to the back - bending evolution brought in that diagram .
it is precisely this typical first - order behavior that will be reflected in the thermodynamical quantities , such as the pressure and densities , as figs .
[ fig4 ] and [ fig5 ] show .
this behavior is responsible for the fact that there is a ( positive ) deviation @xmath113 , which then in turn leads to the surface tension .
we now turn to our numerical results for the surface tension @xmath12 . to this end
we need to determine the free energy density @xmath32 , which requires the evaluation of @xmath217 and @xmath218 for uniform matter thermodynamically unstable region of the phase diagram . for the considered temperature @xmath5 , the associated density region
is bounded by the two coexistence densities @xmath30 and @xmath31 , for which the chemical potential @xmath6 has the same value , as does the pressure @xmath20 . as the density @xmath4 is increased through the lower mechanically metastable ( nucleation ) region , @xmath6 and @xmath20 rise steadily until the lower spinodal boundary has been reached .
then , as @xmath4 moves through the mechanically unstable ( spinodal ) region , both @xmath6 and @xmath20 decrease until the higher spinodal boundary is reached .
they then increase again as @xmath4 moves through the higher mechanically metastable ( bubble - formation ) region , until they finally regain their original values at @xmath219 .
it is convenient to express the ( net ) baryonic density @xmath4 in units of the nuclear saturation density , @xmath220 . generally ,
as is common practice , we subtract from the pressure any finite value it may have in the vaccum .
let us start with @xmath10 for which the relevant results can be readily obtained by taking the @xmath221 limit in eqs .
( [ omegalsm ] ) and ( [ omeganjl ] ) ( see , _
e.g. _ refs .
figure [ fig4 ] shows the pressure as a function of the degree of compression @xmath222 obtained with the njl model for both parameter sets i and ii ; the latter has a stronger coupling and a larger coexistence region .
a qualitatively similar behavior is observed in fig .
[ fig5 ] which shows @xmath6 as a function of @xmath222 for the njl model with both parameter sets .
figure [ fig6 ] shows the behavior of the free energy @xmath223 and its corresponding maxwell line @xmath224 for parameter set ii . in the right panel of fig .
[ fig6 ] we display the difference between these two free energies , @xmath225 for both parameter sets ; this is the key quantity for the determination of the surface tension .
the lsm and the three - flavor njl model yield similar results . at temperatures below criticality , @xmath23 , the thermodynamical potential has two degenerate minima determining the densities of the two coexisting phases , @xmath30 and @xmath31 . in all cases studied here , the lower coexistence density vanishes , @xmath226 .
as for the higher coexistence density , the lsm yields @xmath227 , the two - flavor njl model yields @xmath228 with set
i and @xmath229 with set ii , while the three - flavor njl model gives @xmath230 .
it clear from this figure that set i produces a much weaker phase transition because @xmath113 is much smaller that for set ii , as is indeed reflected in the @xmath12 values shown in table i. in fact , set ii produces a greater coexistence region ( @xmath231 , @xmath232,@xmath233 ) when compared to set i ( @xmath234 , @xmath235 , @xmath236 ) , which is in accordance with the well known fact that set ii should cause the size of the first - order transition line to be longer than the one produced by set i which has a weaker coupling .
further refinements , such as finite-@xmath237 corrections @xcite , contributions from thermal flucutations , and the inclusion of a repulsive vector interaction @xcite , also tend to shrink the first - order transition line @xcite so that , within a fixed parameter set , one should expect these effects to reduce @xmath12 .
table i summarizes all our results for @xmath238 and also lists the characteristic values @xmath53 and @xmath52 as well as the location of the critical point @xmath239 .
the table also provides information related to thermodynamic potential at @xmath10 and @xmath240 by showing the values of the constituent quark mass in vacuum ( @xmath241 ) , which is related to the distance from the global minimum to the origin , as well as the bag constant which gives the energy difference between the local maximum and the global minimum of the potential in vacuum ; these values were taken from refs .
. .[tabfit ] summary of inputs and results .
the length parameter was taken as @xmath242 .
the zero - temperature bag constant @xmath243 and the characteristic energy density @xmath53 are given in @xmath244 , while the effective quark masses in vacuum @xmath245 as well as the critical values @xmath246 and @xmath13 are in @xmath247 .
the resulting zero - temperature surface tension @xmath238 ( given in @xmath248 ) may be compared with the value @xmath249 obtained in ref .
@xcite ) with the lsm in the thin - wall approximation . [ cols="<,>,<,<,<,<,<,<,<",options="header " , ] we finally note that palhares _
et al . _
@xcite , using the approximation @xmath250 obtained the estimate @xmath251 which is very close to our lsm value of @xmath252 and also rather similar to the value @xmath253 resulting from evaluating that intergal using our lsm profile function @xmath254 .
one can easily consider finite temperatures within the employed models .
the interface tension is expected to decrease with increasing temperature because both the coexistence densities @xmath47 and the associated free energy densities @xmath49 move closer together at higher @xmath5 ; they ultimately coincide at @xmath13 where , therefore , the tension vanishes .
this general behavior is confirmed by our calculations , as shown in fig .
the lsm , the njl with parameter set i , and the three - flavor njl display similar behaviors .
the temperature dependence of the surface tension may be relevant for the thermal formation of quark droplets in cold hadronic matter found in `` hot '' protoneutron stars whose temperatures , @xmath255 , are of the order 1020 mev @xcite .
for @xmath255 the relevant value of @xmath256 may be estimated by using table i together with fig .
for example , the three - flavor njl model yields @xmath257 . the temperature dependence of the surface tension is also important in the context of heavy - ion collisions , because it determines the favored size of the clumping caused by the action of spinodal instabilities as the expanding matter traverses the unstable phase - coexistence region
in this work we have shown that the interface tension related to a first - order phase transition may be evaluated once the uniform - matter equation of state is available for the unstable regions of the phase diagram .
it is a convenient feature of the method employed that knowledge of the interface profile functions is is not required , because their determination can be quite complicated , as is the case for njl model @xcite ( although it is easy for the lsm ) .
in addition to the eos , the geometrical approach also requires a proper setting of three input parameters , namely the characteristic densities @xmath52 and @xmath53 together with the length scale @xmath54 .
while this does encumber the numerical results with some degree of uncertainty , our zero - temperature lsm result , @xmath258 , agrees within a few percent with the approximate value obtained in ref .
@xcite , thus suggesting that those parameters were chosen reasonably .
the surface tension determined in the present fashion is entirely consistent with the employed model , including the approximations and parametrizations adopted . for the effective quark models employed here ,
this amounts to considering _ all _ the solutions to the gap equation ( stable , metastable and unstable ) and determine the relevant effective quark masses . in most non - perturbative approximations ( large @xmath237 , mean field , _ etc . _ ) the various quantities of interest , such as the free energy density , become functions of this effective mass and will therefore also reflect the metastable and unstable character of the configuration considered . as a cross check on our procedure
, we have evaluated @xmath238 for the lsm obtaining a result that differs by only about @xmath259 from estimates based on the thin - wall approximation @xcite .
we have investigated the two - flavor njl model as well as its more realistic three - flavor version .
our main conclusion is that all these effective models generate relatively low values for the the surface tension .
this would favor the formation of quark matter and may thus have important astrophysical consequences regarding the existence of pure quark stars . of particular interest is the three - flavor njl result , @xmath260 , because this model is widely used in studies related to neutron stars . here , for simplicity , we have considered pure quark matter where all flavors share the same chemical potential , but it is just a technical matter to generalize our procedure so as to include leptons ( @xmath261 ) in order to enforce @xmath262 equilibrium ( @xmath263 ) , although the additional chemical potential introduces an increased degree of complexity into the features of the phase transition . in principle ,
more refined treatments , such as the polyakov - njl model , can also be considered within the same framework . however , because the effects of the polyakov loop become more important above @xmath264 @xcite we believe that our results , especially the three - flavor ones , can be considered as reasonably accurate , although numerical variations may arise due to the parametrizations and approximations adopted .
mbp was partially supported by cnpq - brazil , while vk and jr were supported by the office of nuclear physics in the us department of energy s office of science under contract no .
de - ac02 - 05ch11231 .
we thank eduardo fraga , sidney avancini , and dbora menezes for their comments and suggestions .
vk would like to thank the departamento de fisica , universidade federal de santa catarina for the kind hospitality during the completion of this work .
|
the surface tension of quark matter plays a crucial role for the possibility of quark matter nucleation during the formation of compact stellar objects , because it determines the nucleation rate and the associated critical size .
however , this quantity is not well known and the theoretical estimates fall within a wide range , @xmath0 .
we show here that once the equation of state is available one may use a geometrical approach to obtain a numerical value for the surface tension that is consistent with the model approximations adopted .
we illustrate this method within the two - flavor linear @xmath1 model and the nambu jona - lasinio model with two and three flavors . treating these models in the mean - field approximation
, we find @xmath2 .
such a relatively small surface tension would favor the formation of quark stars and may thus have significant astrophysical implications .
we also investigate how the surface tension decreases towards zero as the temperature is raised from zero to its critical value .
|
while it is clear that young stars are clustered hierarchically , with ob associations and clusters inside complexes and superassociations , the extent to which there are universal clustering scales is controversial . historically , typical sizes assigned to stellar groupings have depended at least partially on selection criteria and image quality .
efremov ( 1995 ) provides a summary of this history , and concludes that it make sense to consider two distinct scales within the hierarchy : elementary cells on scales of 10 - 100 pc that contain stars with ages up to 10 myr , and large complexes on scales of several hundred parsecs that contain stars with ages up to about 100 myr .
the identification of clusters on different scales is an area of active research , using a variety of surveys and instruments .
for example , livanou et al . ( 2007a , 2007b ) combine optical and infrared obserations to find clusters in the magellanic clouds on scales ranging from 150 pc to 1500 pc , that they classify as aggregates , complexes , or supercomplexes based on size .
in fact , many studies support an approximately scale - free clustering pattern for both the gas and stars in galaxies , a behavior typically attributed to the turbulent cascade of energy within the gas ( see elmegreen & scalo 2004 for a review ) . a quantitative description of stellar clustering including the identification of any characteristic scales
is a key part of several current problems in the formation and evolution of stars , including how closely young stars follow the distribution of gas and dust , how quickly they disperse , and under what conditions they remain gravitationally bound ( e.g. allen et al .
2007 , shu et al .
2007 , yang et al .
2007 , muoz et al .
2007 , pellerin et al .
2007 , carlson et al . 2007 ) .
more generally , it addresses the circumstances under which we should describe young stars in terms of clusters at all .
one way to quantify clustering is through multi - scale variants of cluster identification methods ( e.g. battinelli & efremov 1999 , elmegreen et al .
for example , bastian et al .
( 2007 ) studied the clustering pattern of ob stars in m33 using a multi - scale implementation of the minimum spanning tree method , and found no characteristic scales .
power spectra and correlation functions , on the other hand while unsuited for identifying _ individual _ clusters are more directly geared to searching for special scales , and have been used for this purpose on the integrated light of galaxies ( e.g. elmegreen et al . 2003a ; elmegreen et al .
2003b ; willet et al .
two challenges in interpreting these studies are the fact that integrated starlight includes stars that are older and less clustered , and that individual bright stars create a form of contamination to the power spectrum that can mimic a transition in clustering strength ( elmegreen et al .
odekon ( 2006 ) and parodi & binggeli ( 2003 ) used the integrated autocorrelation function to quantify the clustering strength of individual young stars and individual blue clumps of stars , respectively , in dwarf galaxies , but the number of objects was too small ( 32 - 125 young stars per galaxy , and up to 372 blue clumps per galaxy ) to provide a statistically powerful test for changes in clustering strength with scale . the magellanic clouds photometric survey ( mcps ; zaritsky et al .
1997 ) and massey local group survey ( massey et al . 2006 ; massey et al .
2007 ) provide public catalogs of deep single - star photometry with coverage across the entire star - forming region of most local group galaxies .
single - star photometry allows the separation of clustered , young stars from dynamically relaxed , older stars .
it also prevents the problem of a dominant signal from individual bright stars , because each star counts equally .
four galaxies ( the smc , the lmc , m33 , and m31 ) contain a large number of bright , young stars , enough to provide a statistically powerful test of clustering strength over a wide range of scales .
indeed , harris & zaritsky ( 1999 ) calculated the spatial correlation function for stars in a subsection of the lmc , and found that it is consistent with an exponential function that changes systematically with the mean age of the stellar populations ; however , they were limited to a region 1/18 of the final survey area .
table 1 summarizes the data .
the mcps catalogs give _ ubvi _
stellar photometry for the smc and the lmc down to an apparent magnitude of 21 ( an absolute magnitude of @xmath0 at the distances of the magellanic clouds ) .
they include over five million stars for the smc and over 24 million stars for the lmc .
the _ ubvri _
massey catalogs reach an apparent magnitude of 23 ( an absolute magnitude of about @xmath1 at the distance of m31 and m33 ) with photometric errors less than 10% , and includes 371,781 stars for m31 and 146,622 stars for m33 .
the massey catalogs include the magellanic clouds , but their coverage is not as complete as the mcps .
we select young stars with @xmath2 colors between @xmath3 and 0.0 and absolute magnitudes between @xmath4 and @xmath5 ; this region of the color - magnitude diagram the tip of the main sequence is small enough to minimize foreground contamination ( massey et al .
2006 ; zaritsky et al .
2002 ) , and includes stars younger than about 30 myr ( girardi et al .
we also tried selecting color by @xmath6 and the reddening - free parameter @xmath7 , without a significant change in our results for the correlation function . by eye , the spatial distributions of the bright main sequence stars ( figure 1 ) exhibit clustering over a wide variety of scales .
the data for m31 , a large and highly - inclined galaxy , are probably the most affected by extinction and crowding .
both surveys use multiple scans to cover the star - forming region of each galaxy , and are subject to variations in image quality from scan to scan .
photometric inhomogeneities at scan edges , as well as incompleteness in regions of high crowding and extinction , appear in maps of the locations of fainter stars ( e.g. the red clump stars in figure 5 of zaritsky et al .
2004 ) . in order to assess their effect on features in the correlation function of young stars
, we also calculated the correlation function for red giants , which are slightly fainter than the bright the main sequence stars , and should exhibit a smooth , relaxed distribution .
as described below , this exercise shows that the correlation functions of m33 and m31 are indeed noticeably affected by scan edges , but only on scales larger than the features we discuss here .
to measure clustering as a function of scale we use the angular two - point autocorrrelation function @xmath8 , where @xmath9 is the average number of stars with angular distances between @xmath10 and @xmath11 from any particular star .
the correlation integral @xmath12 and correlation dimension @xmath13 are closely related to this function : @xmath14 , where @xmath15 . for a scale - free distribution ,
both @xmath8 and @xmath12 have the form of a power law , where the power of @xmath12 is greater by two .
for example , the number of points within a distance @xmath16 on a flat , two - dimensional surface covered by a random ( poisson ) distribution of points scales as @xmath17 , yielding a power of two for the correlation integral @xmath18 and a power of zero for @xmath19 . for a distribution that is
more highly clustered , the power law exponent is lower . for a distribution that
not perfectly scale - free , the power law changes with scale . if the goal is to search for such changes , it is preferable to use the derivative form , as we do , rather than the less - sensitive correlation integral . to test the ability of this function to measure clustering strength and find preferred scales , we calculated it for several finite distributions of points ( figure 2 ) . on small scales , the expected power ( the slope on these log - log plots )
is indeed recovered : zero for poisson distributions and @xmath20 ( @xmath21 ) for the sierpinski triangle , a well - known fractal of dimension 1.585 . on the largest scales ,
the finite size of the distribution creates a smooth cutoff . for models with clumps and rings
, the correlation function picks out the sizes of these features and the distances between them .
the area of an annulus with radius @xmath16 is not exactly proportional to @xmath16 because of its nonzero width @xmath22 .
specifically , it scales as @xmath23 , producing a power of @xmath24 instead of one . in our analysis
, we use an increment of 0001 ( 00001 for m33 , which has more resolved stars in the sample ) , so the discrepancy between one and @xmath24 is less than 10% on scales larger than 001 ( 0001 for m33 ) .
a still smaller correction comes from the fact that angular distances are projected onto the sphere of the sky , so that the circumference scales as @xmath25 instead of @xmath10 ; for scales smaller than 5 this produces a difference in the slope of less than 1%
figure 3 shows the correlation functions for the brightest main sequence stars in each galaxy and , for comparison , the bright red giants .
the correlation functions for the red giants are similar to those for the poisson disk and ellipsoid flat at small scales with a smooth cutoff at large scales .
an exception is the slight decrease at about a tenth of a degree in m31 and m32 , reflecting the inhomogeneities at scan edges described in section 2 .
the correlation function for young stars is different from the red giants in two important ways : its slope is lower , indicating a more highly clustered distribution , and it exhibits features at about one kpc for the lmc and m31 and at about 300 pc for m33 . for the lmc ,
a slight decrease in slope precedes a brief flattening , a behavior qualitatively similar to that for clumps and rings . for m33 and m31 , only a transition to a flatter slope is evident . on scales below the transition ,
the power - law exponents for the young stars are -0.2 , -0.4 , -0.6 , and -1.0 ( that is , correlation dimensions of 1.8 , 1.6 , 1.4 , and 1.0 ) for the smc , lmc , m33 , and m31 , respectively . splitting the stars into sub - populations by location on the sky produces correlation functions qualitatively similar to those for the full populations in particular , the existence or non - existence of the transition features remain unchanged .
the largest difference is for the lmc , where the eastern half produces a peak at the approximate radius of the two large clumps on that side ( see figure 1 ) , but both sides still show a feature at about one kpc . splitting the stars into sub - populations by color or magnitude produces steeper power laws for bluer and brighter stars .
the largest difference can be dectected in m33 , where the power is as steep as @xmath26 for main sequence stars with @xmath27 and @xmath28 but only @xmath29 for main sequence stars with @xmath30 and @xmath31 ( figure 4 ) .
the primary cause for this trend is probably a younger average age for the brighter and bluer stars . indeed , many types of measurements have suggested a gradual decrease in clustering with age , presumably illustrating the dissipation of clusters with time ( e.g. harris & zaritsky 1999 , zhang et al .
2001 , schmeja & klessen 2006 , odekon 2006 , de la fuente marcos & de la fuente marcos 2006 , snchez et al . 2007 )
. this trend could also reflect mass segregation , with the bluer and brighter stars more clustered because of a higher average mass rather than a younger average age .
in addition , the brighter stars may have more complete photometry that allows us to see the more intense clustering on smaller scales , although it seems unlikely that this would make a difference on scales as large as hundreds of parsecs .
we tried selecting stars of an intermediate age in each galaxy , considerably further down the main sequence at @xmath32 . for the magellanic clouds ,
the results were similar to those for the brighter main sequence stars , but slightly less clustered .
for m33 and m31 , this population is significantly polluted with randomly - scattered foreground stars , and is affected by the inhomogeneities at scan edges .
figure 4 includes this population for m33 . for both the lmc and the smc
, we also computed the correlation function using data from the massey catalogs , which have different spatial coverage ; these gave the same results as the mcps .
the correlation function for all four galaxies is scale - free over about two orders of magnitude , starting from the smallest scales we can resolve with these data : about 20 pc in m31 , and just a few parsecs in the other galaxies .
a transition feature in the correlation function appears at one kpc in the lmc and m31 and at 300 pc in m33 , above which the slope flattens .
this scale roughly corresponds to the fundamental complexes of efremov ( 1995 ) and the size elmegreen ( 2006 ) called the largest globular regions of star formation in galaxies .
the transition in the lmc occurs at a considerable fraction of the galaxy size about 16% .
only a small number of clumps at this size or larger could exist in the galaxy , and their properties drive the appearance of the transition . for m33 , on the other hand , the transition scale is small enough relative to the galaxy about 3% that the absence of features on larger scales is statistically meaningful .
the transition features are not caused by simple edge or boundary effects .
in particular , they do not represent the geometry of the galaxies as projected on the sky . for a projected distribution with a decreasing density profile ( e.g. the first two models of figure 1 ) , the slope steepens .
similarly , the scan inhomogeneities apparent in the red giant correlation functions for m33 and m31 create a steepening slope , a mini - edge .
the transition features in the data , on the other hand , are an abrupt _ flattening _ on large scales .
we suggest a different connection between the transition scale and disk geometry . at the transition scale ,
density fluctuation modes become highly dependent on disk structure . from a dynamical standpoint ,
the modes that develop within a rotating disk depend partly on the disk thickness ( e.g. tasker & bryan 2008 ) . and of course , spherical clusters ( elmegreen s `` globular regions '' ) that are too large can not exist inside the two - dimensional disk . padoan et al .
( 2001 ) and elmegreen et al .
( 2001 ) found that the clustering of hi in the lmc shows a feature at about 180 pc that they identify as the transition from small - scale three - dimensional turbulence to large - scale two - dimensional disturbances in the disk .
more recently , kim & park ( 2007 ) found a transition in hi for the lmc at 290 pc .
we may be seeing the analogous transition for stars .
following efremov ( 1995 ) , we might expect a smaller characteristic scale for the youngest stars and a larger one for slightly older stars .
however , when we subdivide the young stars by magnitude or color , we see that the location of the transition changes only slightly , if at all , even when there is a considerable change in slope .
figure 4 shows the most extreme case , where the bluest and brightest subpopulation of main sequence stars in m33 have a considerably steeper slope , but no characteristic scales smaller than the transtion scale at several hundred parsecs .
it is true that the correlation function for m31 does not reach scales even as small as 15 pc . a combination of photometry from multiple _ hubble space telescope _ pointings ( e.g. those used by williams & hodge 2001 to identify clusters in m31 on a scale of about five pc ) could provide a statistically meaningful test on this scale .
to address this question further , it would be useful to select stars with a greater variety of age categories , including very young populations as well as intermediate - aged populations that are not highly contaminated by foreground stars .
possible approaches to this include infrared identification of young stellar objects like those in bolatto et al .
( 2007 ) , spectroscopic identification , and selection of cepheids with a specific period .
another improvement would be to use results from population synthesis modeling of an entire color - magnitude diagram to restrict the ages that are statistically dominant in a certain regions of the bright main sequence . a related question , one that could also be addressed using a broader range of scales and age categories , is the statistical detection of cluster dissipation with age , an effect that might cause an additional transition , at a scale that increases with age , to _ weaker _ clustering on small scales .
mass segregation would probably have a very similar effect .
in both the lmc and m33 we see a hint of this behavior in the correlation dimension for the fainter subset of main sequence stars , but the correlation function is statistically messy on these scales , with this smaller sub - populations of stars . on scales below the transition ,
the correlation dimensions for young massive stars varies from about 1.0 for m31 and the brightest subpopulation of m33 to 1.8 for the smc .
the variation in clustering strength may reflect differences in the dominant ages or masses of stars , but a quantitative understanding of the correlation dimension really requires a detailed assessment of crowding and extinction , beyond the scope of this paper . crowding and extinction are likely to cause increasing incompleteness on smaller scales , even for stellar populations that are completely recovered in simple artificial - star tests ( see the discussion , for example , in odekon 2006 ) .
this is likely to be an especially large problem for m31 , which is highly inclined as well as large and dusty . on the other hand , as a rough completeness check in the smc , we compared the mcps photometry in the cluster ngc 346 with the high - resolution advanced camera for surveys photometry of that region ( gouliermis et al .
they show excellent agreement , the only missing stars residing in a tiny cluster about one pc across .
these correlation dimensions are based on two - dimensional , projected distributions . to estimate the true three - dimensional value ,
it is necessary to consider the extent to which the observations represent projections rather than slices or outer skins of each galaxy .
while a slice through a fractal gives a dimension less by one , a projection of a fractal approximately reproduces the three - dimensional value if it is less than two , and gives two otherwise ( see snchez et al .
2005 for detailed models ) .
for this reason , galaxies that are more transparent may appear to have a larger two - dimensional correlation dimension , simply because we see their distribution more as a projection than as a thin , outer skin .
this is probably not , however , the only reason for the higher dimension we observe in the smc compared with m31 , given the systematically lower correlation dimensions for younger subsets of stars , as expected for populations relaxing with age . finally , our results show that it is not surprising to see a fairly wide range of correlation dimensions from galaxy to galaxy , ( e.g. those in parodi & binggeli 2003 and odekon 2006 ) , given the existence of a transition scale in some galaxies and the differences in clustering strength for different population of bright main sequence stars .
this project relies on data products from the magellanic clouds photometric survey of zaritsky et al .
( 2002 , 2004 ) , accessible at http://ngala.as.arizona.edu/dennis/mcsurvey.html , and from the local group survey of massey et al .
( 2006 , 2007 ) , accessible at + http://www.lowell.edu/users/massey/lgsurvey.html .
allen , l. , et al .
2007 , in protostars and planets , ed .
v , b. reipurth , d. jewitt , & k. keil ( tucson : university of arizona press ) , 361 bastian , n. , ercolano , b. , gieles , m. , rosolowsky , e. , scheepmaker , r.a . , gutermuth , r. , efremov , yu .
2007 , , 379 , 1302 battinelli , p. & efremov , yu . n. 1999 , a&a , 346 , 778 bolatto , a. d. et al .
2007 , , 655 , 212 carlson , l. r. et al .
2007 , , 665 , l109 de la fuente marcos , r. & de law fuente marcos , c. 2006 , , 372 , 279 efremov , yu .
n. 1995 , , 110 , 2757 elmegreen , b. g. 2006 , in the many scales of the universe , ed .
e. j. alfaro et al .
( dordrecht : springer ) elmegreen , b. g. , elmegreen , d. m. , chandar , r. , whitmore , b. , regan , m. 2006 , , 644 , 879 elmegreen , b. g. , elmegreen , d. m. , leitner , s. 2003a , , 590 , 271 elmegreen , b. g. , leitner , s. n. , elmegreen , d. m. , & cuillandre , j .- c .
2003b , , 593 , 333 elmegreen , b. g. , & scalo , j. 2004 , , 42 , 211 elmegreen , b. g. , sungeun , k. & staveley - smith , l. 2001 , , 548 , 749 girardi , l. , bertelli , g. , bressan , a. , chiosi , c. , groenewegen , m. a. t. , marigo , p. , salasnich , b. , & weiss , a. 2002 , a&a , 391 , 195 gouliermis , d. a. , dolphin , a. e. , brandner , w. , & henning , t. 2006 , , 166 , 549 harris , j. & zaritsky , d. 1999 , , 117 , 2831 kim , s. , & park , c. 2007 , , 663 , 244 livanou , e. , gonidakis , i. , kontizas , e. , klein , u. , kontizas , m. , kester , d. , fukui , y. , mizuno , n. , & tsalmantza , p. 2007a
, , 133 , 2179 livanou , e. , kontizas , m. , gonidakis , i. , kontizas , e. , maragoudaki , f. , oliver , s. , efstathiou , a. , & klein , u. 2007b , arxiv : astro - ph/0512193v3 massey , p. , olsen , k. a. g. , hodge , p. w. , jacoby , g. h. , mcneill , r. t. , smith , r. c. , & strong , s. b. 2007 , , , 133 , 2393 massey , p. , olsen , k. a. g. , hodge , p. w. , strong , s. b. , jacoby , g. h. , schlingman , w. , & smith , r.c .
2006 , , 131 , 2478 munoz , d. j. , mardones , d. , garay , g. , rebolledo , d. , brooks , k. , & bontemps , s. 2007 , , 668 , 906 odekon , m. c. 2006 , , 132 , 1834 padoan , p. , kim , s. , goodman , a. , & staveley - smith , l. 2001 , , 555 , l33 parodi , b. r. , binggeli , b. , 2003 , a&a , 398 , 501 pellerin , a. , meyer , m. , harris , j. , & calzetti , d. 2007 , , 658 , l87 snchez , n. , alfaro , e. j. , & prez , e. 2005 , , 625 , 849 schmeja , s. & klessen , r. s. , 2006 , a&a , 449 , 151 shu , f. h. , allen , r. j. , lizano , s. , & galli , d. 2007 , , 662 , l75 tasker , e. j. , & bryan , g. l. 2008 , , 673 , 810 willett , k. w. , elmegreen , b. g. , & hunter , d. a. 2005 , , 129 , 2186 williams , b. f. , & hodge , p. w. 2001 , , 559 , 851 yang , c .- c . ,
gruendl , r. a. , chu , y .- h . ,
mac low , m .- m . , & fufui , y. 2007 , , 671 , 374 zaritsky , d. , harris , j. , & thompson , i. 1997 , , 114 , 1002 zaritsky , d. , harris , j. , thompson , i.b .
, & grebel , e. k. 2004 , , 128 , 1606 zaritsky , d. , harris , j. , thompson , i.b . , grebel , e. k. , & massey , p. 2002
, , 123 , 855 zhang , z. , fall , s. m. , & whitemore , b. c. 2001 , , 561 , 727 clrccl galaxy&type & @xmath33 & distance modulus & number of & reference + & & & @xmath34 & selected stars & + smc & ir iv - v & -17.1 & 19.0 & 798 & zaritsky et al .
2002 + lmc & ir iii - iv & -18.5 & 18.5 & 2332 & zaritsky et al .
2004 + m33 & sb i - ii & -18.9 & 24.7 & 3796 & massey et al . 2006 + m31 & sc ii - iii & -21.2 & 24.5 & 1188 & massey et al .
|
the autocorrelation function provides an objective test for the existence of special scales in the hierarchical clustering of young stars .
we apply this measure to single - star photometry for the brightest main sequence stars in the small magellanic cloud ( smc ) , the large magellanic cloud ( lmc ) , m33 , and m31 , using data from the magellanic clouds photometric survey and the massey local group survey .
our primary result is the identification of a transition to a higher correlation dimension ( weaker clustering ) at one kpc in the lmc and m31 , and at 300 pc in m33 .
we suggest that this transition marks the large - scale regime where disk geometry and dynamics set the scale for structure . on smaller scales ,
the correlation functions for each galaxy are scale - free over at least two orders of magnitude , with a projected correlation dimension varying from 1.0 for m31 to 1.8 for the smc .
this variation is probably caused by a combination of differences in stellar ages and masses , physical environment , and extinction .
|
the cuo@xmath4 plaquette is the key structural element in the majority of low - dimensional copper oxides @xcite . ideally , it consists of the cu atom at the center of the square formed by four oxygen atoms .
the valence 2 + of the cu ion corresponds to the atomic configuration in which all @xmath8 orbitals are fully occupied except that of the @xmath9 symmetry containing one unpaired electron and having the highest energy .
such a geometry leads to a strong hybridization between the @xmath9 states of cu and @xmath10 states of oxygen , which affects magnetic properties of low - dimensional cuprates .
for instance , spin density is largely delocalized and spread over the plaquette .
this effect should be taken into account when analyzing the data from neutron scattering @xcite .
the mutual orientation of the cuo@xmath4 plaquettes in a particular system is very important and responsible for nontrivial magnetic and electronic properties observed experimentally .
depending on the connectivity of the plaquettes , the interaction between magnetic moments can be ferromagnetic or antiferromagnetic .
here , the copper - oxygen - copper bond angle plays crucial role .
the topology of ba@xmath1cu@xmath1sc@xmath2o@xmath3 and isostructural ba@xmath11cu@xmath11in@xmath4o@xmath3 is a distinct one , since it contains three mutually orthogonal plaquette sublattices presented in fig .
[ im : crystal ] .
this type of geometry was coined a paper - chain model @xcite . within each paper chain ,
the neighboring cuo@xmath4 plaquettes have one common oxygen site .
the corresponding angle of the cu
o cu bond is close to 90@xmath12 , which means that kinetic superexchange processes between the neighboring cu sites in ba@xmath1cu@xmath1sc@xmath2o@xmath3 are strongly suppressed .
in this situation , other types of the magnetic couplings may become important .
for example , the absence of the inversion symmetry between nearest - neighbor cu sites in ba@xmath1cu@xmath1sc@xmath2o@xmath3 results in the antisymmetric anisotropic exchange interaction that , when taken on its own , will favor orthogonal spin configuration proposed as a candidate ground state of ba@xmath1cu@xmath1in@xmath2o@xmath3 @xcite . on the other hand
, the non - zero curie - weiss temperature @xmath13 = -70k estimated from the high - temperature magnetic susceptibility indicates that isotropic exchange couplings are non - negligible and ferromagnetic .
( right panel ) paper chains formed by cuo@xmath4 plaquettes .
barium , scandium and interchain oxygen atoms are excluded for clarity .
crystal structures are visualized by using the vesta software @xcite.,title="fig:",scaledwidth=23.0% ] cu@xmath1sc@xmath2o@xmath3 .
( right panel ) paper chains formed by cuo@xmath4 plaquettes .
barium , scandium and interchain oxygen atoms are excluded for clarity .
crystal structures are visualized by using the vesta software @xcite.,title="fig:",scaledwidth=23.0% ] consecutive experimental and theoretical studies @xcite revealed that intrachain couplings are strong and ferromagnetic , while interchain couplings are weak and antiferromagnetic
. however , no conclusive information on the magnetic ground state has been reported , and the relevance of frustrated next - nearest - neighbor couplings within the paper chain remained controversial .
last but not least , anisotropic exchange interactions were disregarded in previous theoretical studies , despite the fact that the orthogonal geometry of cuo@xmath4 plaquettes should favor such contributions to the exchange . in the following , we address all these questions in a comprehensive study combining first - principles calculations and accurate numerical simulations of both magnetic ground state and finite - temperature properties .
we show that the magnetic ground state of ba@xmath1cu@xmath1sc@xmath2o@xmath3 is collinear , with ferromagnetic order within the paper chains and antiferromagnetic order between the chains .
magnetic anisotropy determines the direction of the magnetic moment without affecting the ground state qualitatively .
_ structural properties_. the polycrystalline sample of ba@xmath11cu@xmath11sc@xmath4o@xmath3 was prepared at high temperatures via solid state reaction from the stoichiometric mixture of high - purity baco@xmath11 , cuo , and sc@xmath14o@xmath11 .
these reagents were ground , pelletized and fired in alumina crucibles at 850 950 @xmath12c in air for 3 days with regrinding every 24 h. then the sample was quenched in air to room temperature .
phase purity of the sample was confirmed by powder x - ray diffraction data taken using the radian-2
diffractometer with cuk@xmath15 radiation over a @xmath16 range of @xmath17 . the crystal structure of ba@xmath1cu@xmath1sc@xmath2o@xmath3 presented in fig .
[ im : crystal ] has the tetragonal @xmath18 space group with lattice parameters @xmath19 , @xmath20 , @xmath21@xmath22 , @xmath23 @xcite .
the main structural units of the ba@xmath1cu@xmath1sc@xmath2o@xmath3 compound are cuo@xmath4 plaquettes forming a chain along along the @xmath24 axis . within an individual chain , there are two nonequivalent crystallographic positions for copper , cu@xmath25 and cu@xmath26 in the 1:2 ratio .
the copper - oxygen plaquette for cu@xmath25 is undistorted and located in the @xmath27 plane .
the distances between copper and oxygen atoms are 2.028 .
the situation is different for cu@xmath26 .
the plaquette for this copper atom is orthogonal to the @xmath27 plane and compressed along the @xmath24-axis .
the copper - oxygen distance for the cu@xmath26 atom varies from 1.96 to 2.018 .
the plaquettes in the chain are rotated by @xmath28 around the @xmath24 axis with respect to the plaquettes belonging to the neighboring chains .
barium and scandium atoms are located between the chains . as we will show below
, scandium orbitals participate in the formation of exchange paths between cu@xmath29 ions in ba@xmath1cu@xmath1sc@xmath2o@xmath3 .
some of the oxygen atoms form bonds between the chains through scandium atoms .
these bonds and interactions play significant role in the formation of long - range magnetic order in this system . _ magnetic properties .
_ field and temperature dependencies of the magnetization in ba@xmath11cu@xmath11sc@xmath4o@xmath3 were measured in the temperature range @xmath30 k and in magnetic fields up to 9 t by means of a physical properties measurement system ( ppms9 t ) from quantum design .
temperature dependence of the magnetic susceptibility @xmath31 measured in the applied field @xmath32 t is represented in fig .
[ im : thermodynamics ] ( left panel ) . at high temperatures ,
the susceptibility follows the curie - weiss behavior ( insert on the left panel of fig.[im : thermodynamics ] ) @xmath33 with the temperature - independent part @xmath34 ( emu / mol ) , @xmath35k and @xmath36 ( emuk / mol ) . from the value of the curie
constant , the @xmath37-factor can be estimated as 2.05 . of ba@xmath1cu@xmath1sc@xmath2o@xmath3 measured in the applied field
@xmath32 t .
the inset shows the curie - weiss approximation with parameters @xmath35k and @xmath38(emuk / mol ) , as denoted by the red lines .
( right panel ) field dependence of the magnetization measured at different temperatures above and below @xmath39
. , scaledwidth=50.0% ] the negative value of the curie - weiss temperature @xmath40k suggests that the leading exchange interactions in ba@xmath1cu@xmath1sc@xmath2o@xmath3 are ferromagnetic , but below 180k there is a strong negative deviation from the curie - weiss behavior .
this deviation indicates the increase of antiferromagnetic correlations , which eventually yield antiferromagnetic long - range order below @xmath41k . as a consequence
, one can observe a sharp peak in the temperature dependence of the magnetic susceptibility @xmath31 .
field dependence of the magnetization measured at different temperatures is shown in fig .
[ im : thermodynamics ] ( right panel ) .
the value of the saturation moment at 3k equals to 2.89@xmath42 , which roughly corresponds to three cu@xmath29 ions with the local spin-1/2 .
importantly , there is a deviation of the @xmath43 curve from the linear dependence for the external magnetic fields smaller than 3 t and at temperatures below @xmath44 .
the same feature of the magnetization curve was found for the ba@xmath1cu@xmath1in@xmath2o@xmath3 compound @xcite . as we will show below
, this peculiarity is due to the inter - atomic symmetric anisotropic exchange interactions .
to describe the electronic structure and magnetic properties of the ba@xmath1cu@xmath1sc@xmath2o@xmath3 compound , we performed calculations within density functional theory ( dft ) @xcite using the generalized gradient approximation ( gga ) with the pbe exchange - correlation functional @xcite . to this end , we employed quantum espresso @xcite and vienna ab initio simulation package ( vasp ) @xcite codes . in these calculations
, we set an energy cutoff in the plane - wave construction to 400ev and the energy convergence criteria to 10@xmath45ev .
for the brillouin - zone integration , a 6@xmath466@xmath466 monkhorst - pack mesh was used .
magnetic behavior of ba@xmath1cu@xmath1sc@xmath2o@xmath3 is described by the following spin hamiltonian : @xmath47 \nonumber \\
+ \mu_{b}\sum\limits_{i}\hat{\mathbf{s}}_i \tensor g_i\mathbf{b } , \label{eq : magnetic_model}\end{aligned}\ ] ] where @xmath48 , @xmath49 and @xmath50 are isotropic , antisymmetric anisotropic and symmetric anisotropic exchange interactions between spins , respectively .
@xmath51 is the g - tensor for @xmath52 site , and @xmath53 is the external magnetic field .
exchange interaction parameters can be estimated by different methods .
for instance , they can be calculated by using superexchange theory @xcite on the basis of the dft results . the corresponding expression for the _ isotropic exchange interaction _ is given by @xmath54 @xcite , where @xmath55 and @xmath56 are the hopping integral and on - site coulomb interaction , respectively .
this expression describes only the antiferromagnetic kinetic contribution to the total magnetic coupling between the two sites . in the case of cuprates with nearly 90@xmath12 metal - ligand - metal bond angle , there is a ferromagnetic contribution that is formed due to the direct overlap of the neighboring wannier functions @xcite .
all contributions to the isotropic exchange interaction can be accounted for on the level of dft+@xmath56 calculations @xcite by computing differences between total energies of collinear magnetic configurations @xcite .
these results will be presented in section iii c. in order to estimate symmetric ( @xmath50 ) and antisymmetric ( @xmath57 ) anisotropic interactions , we used the superexchange theory proposed by moriya @xcite and further developed in ref .
: @xmath58 , \label{eq : dm - vector}\ ] ] @xmath59 , \label{eq : gamma - matrix}\ ] ] where @xmath60 are pauli matrices . within this approach ,
one calculates the hopping integrals @xmath61 taking the spin - orbit coupling into account .
in contrast to the previous investigations @xcite , where such hoppings were estimated by using the perturbation theory on the spin - orbit coupling , we calculate them from first - principles here , thus treating the spin - orbit coupling on the fully _ ab initio _ level . to this end , the dft+so scheme was used . to determine the @xmath37-tensor in eq .
, we use the second - order perturbation theory described in ref . .
considering the wannier function of the @xmath9 symmetry as the ground - state orbital , one obtains @xmath62 in this expression , @xmath63ev is the spin - orbit coupling constant of the copper atom , and @xmath64 where @xmath65 = @xmath66 .
the index @xmath67 runs over the excited states of the copper atom , where one of the @xmath68-orbitals ( @xmath69 , @xmath70 , @xmath71 and @xmath72 ) is half - filled , whereas the @xmath9 orbital is fully filled .
the corresponding orbital energies are defined from dft calculations .
due to the complex geometry of ba@xmath1cu@xmath1sc@xmath2o@xmath3 , in our study we use @xmath73 and @xmath74 notations for local and global coordinate systems , respectively .
the fermi level is at zero energy .
the high - symmetry @xmath75-points in the first brillouin zone are the following : @xmath76 , @xmath77 , @xmath78 .
( right ) the calculated partial densities of states for copper and oxygen atoms.,scaledwidth=44.0% ] _ electronic structure .
_ the calculated gga electronic structure is presented in fig . [ im : bands ] .
there are six well - separated bands at the fermi level .
according to the calculated partial density of states , these bands correspond to strongly hybridized @xmath8 copper and @xmath10 oxygen states . due to a strong overlap of copper orbital of the @xmath9 symmetry and @xmath10 orbitals of the oxygen atoms ,
the complex molecular orbital is formed .
the corresponding bonding and antibonding states are centered at @xmath79ev and 0ev , respectively ( fig .
[ im : partial_dos ] ) . atoms .
the green solid line corresponds to the @xmath9 orbitals , the violet dashed line denotes the overall density of states for the four other orbitals ( @xmath69 , @xmath71 , @xmath80 , @xmath72 ) .
the fermi level is at zero energy .
the spectral functions of cu@xmath26 in the local coordinate system are close to those calculated for cu@xmath25.,scaledwidth=44.0% ] to perform quantitative analysis of the dft results , we constructed a low - energy model by using the maximally localized wannier functions ( mlwf ) procedure , as described in refs . and .
this procedure is implemented in the wannier90 code @xcite .
the example of the constructed wannier orbital is presented in the insert of fig .
[ im : f - factor ] .
one can see that , being centered at the copper atom , the orbital has large tails at the oxygen sites belonging to the plaquette .
the electronic density described by the wannier function centered on the copper atom can be decomposed as follows : 52% on the central copper atom , 11% on each oxygen atom in the plaquette , and approximately 1% on nearest - neighbor copper atoms .
such a delocalization of the electronic density affects the magnetic behavior .
for instance , the account of the hybridization with oxygen states changes the magnetic form factor , which is the fourier transform of the electronic density @xcite . from fig .
[ im : f - factor ] one can see that the covalent form factor decays much faster at small * q * than the pure ionic cu@xmath29 one @xcite . as we will show below
, such a redistribution of the scattering density and the ensuing renormalization of the magnetic form factor plays an important role for simulating neutron scattering spectra .
form factor obtained within the 3-gaussian approximation @xcite and covalent form factor calculated for ba@xmath1cu@xmath1sc@xmath2o@xmath3 by using the constructed wannier orbitals .
the insert shows the wannier orbital of @xmath9 symmetry taken from the mlwf procedure , centered on the cu@xmath25 atom in the cuo@xmath4 plaquette .
the blue and red spheres correspond to copper and oxygen atoms , respectively.,scaledwidth=44.0% ] the hopping matrices between different copper sites in the basis of wannier functions of @xmath9 symmetry are represented in table [ tab : hoppings ] .
they agree reasonably well with those obtained by using the tight - binding linearized muffin - tin orbitals method @xcite .
the difference can be attributed to the fact that the structure of ba@xmath1cu@xmath1sc@xmath2o@xmath3 is not close - packed . in this case
, the tb - lmto method is very sensitive to computational details , such as the radii of muffin - tin spheres and the composition of the valence shell .
we demonstrate this in the supplementary material .
.comparison of the hopping integrals ( inmev ) calculated by using the mlwf method in this work and the downfolding procedure described in ref . .
the interaction paths are labeled in accordance with fig .
[ im : exchange ] ( left ) and additionally denoted by cu
cu distances @xmath81 ( in ) . [ cols="<,<,^,^",options="header " , ] [ tab : isotropic_exchange ]
according to anderson s superexchange theory , the distant intra- and inter - chain interactions @xmath82 and @xmath83 are much smaller than @xmath84 ( table [ tab : isotropic_exchange ] ) .
this scenario is well reproduced by our dft+@xmath56 calculations , in contrast to the dft+@xmath56 results reported in ref . .
moreover , we observe a large difference between the values of antiferromagnetic exchange interactions @xmath84 , @xmath85 , @xmath82 and @xmath83 .
the curie - weiss temperature was estimated by using isotropic exchange interactions , @xmath86 , where @xmath87 is the boltzmann constant .
taking into account the number of atoms of each type in the unit cell , the averaged curie - weiss temperature can be defined as @xmath88 .
for @xmath89=11ev we obtain @xmath90k , which is consistent with the experimental value of @xmath91k .
the account of the spin - orbit coupling within the dft+@xmath56 scheme gives us an opportunity to study orbital magnetism in ba@xmath1cu@xmath1sc@xmath2o@xmath3 .
the size of the orbital magnetic moment on the given copper site in the chain depends on the direction of the total magnetization of the system .
for instance , we get @xmath92 = 0.175 @xmath42 and @xmath93 = 0.045 @xmath42 when the total magnetic moment is along the @xmath24-axis .
thus , the preferable direction for the orbital moment is normal to the cuo@xmath4 plaquette .
this effect can be explained on the level of the crystal - field splitting revealed by the dft calculations ( table [ tab : wannier_energies ] ) .
the energies of the @xmath69 and @xmath71 ( @xmath80 ) orbitals are close to each other . however , the corresponding matrix elements of the orbital momentum operator are very different , @xmath94 = @xmath95 and @xmath96 = @xmath97 . therefore , within the perturbation theory on the spin - orbit coupling , the preferred positioning of the orbital moment along the normal of the plaquette
is confirmed . from dft+@xmath56+so calculations we also estimated diagonal components of the anisotropic @xmath98 by using the total energies difference method .
the obtained diagonal components are @xmath99mev and @xmath100mev .
the symmetry of the resulting @xmath98 agrees with those estimated by using the perturbation theory eq .
( [ eq : gamma - matrix ] ) . however , the absolute values of the @xmath98 elements obtained in dft+@xmath56+so are about five times larger than those from dft+so calculations presented in the previous section .
such a difference can be naturally explained by the fact that the perturbation theory results were obtained for the one - band model with spin - orbit coupling .
the account of the complex multi - orbital structure of the copper atoms within dft+@xmath56+so scheme gives additional contributions to the inter - site anisotropic exchange interaction .
these contributions originate from intra - atomic hund s rule exchange , as discussed in ref . .
in this section , we investigate the magnetic model of ba@xmath1cu@xmath1sc@xmath2o@xmath3 , define the ground - state spin configuration , and describe magnetic properties observed experimentally .
we start with the solution of the magnetic model considering spins as classical vectors .
three magnetic configurations presented in fig . [ im : local_minimum ] are considered : * @xmath101 ferromagnetic chains with spins directed along the @xmath24-axis ( fig .
[ im : local_minimum]a ) ; * @xmath53 ferromagnetic chains with spins directed along the @xmath102 axis ( fig .
[ im : local_minimum]b ) ; * @xmath103 orthogonal spin arrangement ( fig.[im : local_minimum]c ) proposed in ref . . in all three cases , antiferromagnetic ordering between the chains
is assumed .
we found that the configuration @xmath101 has the lowest energy .
the energy of the configuration @xmath53 is 0.028mev higher according to the the difference @xmath104 .
the configuration @xmath103 has the highest energy , because the leading exchange interaction @xmath105 does not contribute to the total energy of this orthogonal spin configuration . since the absolute value of @xmath105 is much larger than @xmath84 and @xmath85 , we can tentatively neglect frustration by second - neighbor interactions within the chain and proceed to the numerical treatment of the quantum spin model by quantum monte carlo ( qmc ) simulations .
thermodynamic properties were obtained using stochastic series expansion ( sse)@xcite method implemented in the loop@xcite algorithm of the alps@xcite simulation package .
.,scaledwidth=44.0% ] we performed simulations for @xmath106 finite lattices with @xmath107 @xmath108 20 and periodic boundary conditions . in these calculations , we take into account the leading intra - chain interaction @xmath105 and inter - chain couplings @xmath83 estimated from dft+@xmath56 . to take other inter - chain couplings into account , an effective interaction of the @xmath109 type between cu@xmath26 atoms
is introduced into our model .
the value of this interaction was estimated by using superexchange theory , @xmath110 = 0.093mev . to include the effects of the spin - orbit coupling
, we used anisotropic exchange interactions between nearest neighbors .
the corresponding @xmath111-tensors taken from the dft+@xmath56+so calculations have diagonal form for all bonds and favor spin alignment along the @xmath24-axis . from fig.[im : susceptibility ] one can see that the magnetic susceptibility calculated using _ ab initio _ microscopic parameters reproduces experimental data quite well . here
, we used the averaged value of the @xmath37-factor , @xmath37 = 2.138 estimated from the first - principles calculations indicates the phase transition at @xmath112k.,scaledwidth=44.0% ] to calculate the nel temperature , we used two complementary approaches proposed in ref . .
they are based on the determination of the spin stiffness and binder s cumulant for staggered magnetization at different temperatures and for different lattice size .
these results are represented in fig .
[ im : phase_transition ] ( top and bottom panels , respectively ) .
the crossing point for finite lattices of different size identifies the nel temperature @xmath112k , which is in good agreement with the experimental value of 16.4k .
since the magnetization data were obtained for polycrystalline sample , we simulated magnetization of our anisotropic spin model for the @xmath24 and @xmath113 directions of the external magnetic field . the powder - averaged magnetization curve is presented in fig .
[ im : magnetization ] . within our approach
we can reproduce the experimental profile of the magnetization at small magnetic fields that is characterized by the deviation from the linear behavior .
this feature of the @xmath43 curve originates from the anisotropic exchange interaction . in the insert of fig .
[ im : magnetization ] , the directions of local magnetic moments in nearest - neighbor chains are shown . upon increasing external magnetic field
, the ground state goes through a spin - flop transition toward collinear ferromagnetic order between the chains .
another important quantity characterizing a low - dimensional spin system is the value of the local magnetic moment . in quantum systems with @xmath114 , it can be considerably reduced from its nominal value of 1@xmath115 . for instance , in the case of the antiferromagnetic square lattice @xcite the value of the magnetic moment equals to 0.6@xmath42 .
it indicates a significant suppression of the magnetic moment by quantum fluctuations in a low - dimensional antiferromagnetic spin system . for a realistic estimation of ordered magnetic moment in ba@xmath11cu@xmath11sc@xmath4o@xmath3
, we used the procedure proposed in ref . .
to this end , the non - local spin - spin correlation function @xmath116 between the most distant spins of a finite lattice is used .
the corresponding value of the magnetic moment per site equals to @xmath117 .
the ordered moment in the thermodynamic limit @xmath118 can be estimated by the extrapolation procedure @xcite : @xmath119 where @xmath120 , @xmath121 and @xmath122 are constants .
because of two inequivalent positions of copper atoms in the ba@xmath1cu@xmath1sc@xmath2o@xmath3 system , we have calculated the spin - spin correlation functions for cu@xmath25 and cu@xmath26 sublattices , respectively .
these simulations produce the local magnetic moments of 1 @xmath115 on both cu sites ( insert of fig .
[ im : susceptibility ] ) . to confirm the proposed magnetic ground state of ba@xmath11cu@xmath11sc@xmath4o@xmath3 , we simulated neutron diffraction data for candidate spin configurations .
experimental results are taken from ref . and represented in fig .
[ im : neutron ] . a weak low - angle magnetic reflection at 11.8@xmath123
is clearly seen at 2k .
this reflection disappears at about 20k , which is above the nel temperature .
it is also suppressed in magnetic fields of about 6 t .
while other magnetic reflections may be present as well , the experimental resolution was not high enough to detect them . for modeling neutron diffraction patterns we used the jana2006 package @xcite .
such simulations were performed for three magnetic configurations presented in fig.[im : local_minimum ] .
the resulting spectra are represented in fig .
[ im : neutron ] .
the comparison of the neutron spectra obtained with ionic and covalent magnetic form factors ( fig.[im : f - factor ] ) has shown that the account of the metal - ligand hybridization decreases the intensity of the magnetic reflection at 11.8@xmath123 . .
the insert demonstrates the effect of the copper - oxygen hybridization on the neutron scattering.,scaledwidth=48.0% ] experimental neutron diffraction data are compatible with both configurations @xmath101 and @xmath53 . from the intensity ratio between the magnetic and nuclear peaks at 11.8@xmath123 and 16.7@xmath12 ,
respectively , model @xmath53 shows a slightly better agreement with the experiment .
we note , however , that the intensity of the magnetic reflection is largely determined by the size of the ordered moment and by the magnetic form - factor .
it is also important that configurations @xmath101 and @xmath53 are very close in energy , and their relative stability may depend on the dft exchange - correlation functional and @xmath89 value . on the other hand ,
the configuration @xmath103 proposed in ref .
can be clearly discarded , because it reveals zero intensity of the magnetic reflection at 11.8@xmath123 .
we have reported a comprehensive microscopic description of electronic and magnetic properties of ba@xmath1cu@xmath1sc@xmath2o@xmath3 , a spin-@xmath0 low - dimensional magnet featuring distinct paper - chain topology of spin chains formed by cuo@xmath4 plaquettes .
we show that interactions within the paper chains are ferromagnetic and induce ferromagnetic order along the chains , which are ordered antiferromagnetically . by means of the superexchange theory and first - principles calculations ,
symmetric and antisymmetic anisotropic exchange interactions were determined .
quantum monte carlo similations for the resulting magnetic model reveal excellent agreement with the experimental nel temperature , magnetic susceptibility , field - dependent magnetization , and elastic neutron scattering data .
we demonstrate minor role of intrachain magnetic frustration and pinpoint fingerprints of the magnetic anisotropy in the non - linear behavior of magnetization as a function of field .
the work of dib , vvm and ivs is supported by the grant program of the russian science foundation 14 - 12 - 00306 .
the numerical simulations were performed on the high - performance cluster of ural federal university .
aat was supported by the federal ministry for education and research through the sofja kovalevskaya award of alexander von humboldt foundation .
40ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1103/physrevb.86.144409 [ * * , ( ) ] link:\doibase 10.1103/physrevb.82.184410 [ * * , ( ) ] link:\doibase 10.1103/physrevb.83.094435 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.98.077202 [ * * , ( ) ] link:\doibase 10.1103/physrevb.92.245113 [ * * , ( ) ] link:\doibase 10.1103/physrevb.85.104420 [ * * , ( ) ] link:\doibase 10.1107/s0021889811038970 [ * * , ( ) ] link:\doibase 10.1088/0953 - 8984/24/23/236001 [ * * , ( ) ] link:\doibase 10.1088/0953 - 8984/24/16/166001 [ * * , ( ) ] link:\doibase 10.1039/b008787o [ * * , ( ) ] link:\doibase 10.1103/physrev.140.a1133 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.77.3865 [ * * , ( ) ] link:\doibase 10.1088/0953 - 8984/21/39/395502 [ * * , ( ) ] link:\doibase 10.1103/physrevb.47.558 [ * * , ( ) ] link:\doibase 10.1103/physrevb.54.11169 [ * * , ( ) ] link:\doibase 10.1103/physrev.115.2 [ * * , ( ) ] link:\doibase 10.1103/physrevb.52.10239 [ * * , ( ) ] link:\doibase 10.1103/physrev.120.91 [ * * , ( ) ] link:\doibase 10.1103/physrevb.75.224408 [ * * , ( ) ] link:\doibase 10.1103/physrevb.78.195110 [ * * , ( ) ] link:\doibase 10.1103/physrevb.44.943 [ * * , ( ) ] link:\doibase 10.1103/physrevb.84.224429 [ * * , ( ) ] @noop ( ) link:\doibase 10.1103/physrevb.56.12847 [ * * , ( ) ] link:\doibase 10.1103/physrevb.65.035109 [ * * , ( ) ] link:\doibase 10.1016/j.cpc.2007.11.016 [ * * , ( ) ] @noop ( ) link:\doibase 10.1103/physrevb.12.3060 [ * * , ( ) ] link:\doibase 10.1103/physrevb.27.7144 [ * * , ( ) ] link:\doibase 10.1103/physrevb.92.235109 [ * * , ( ) ] link:\doibase 10.1103/physrevb.43.5950 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.87.047203 [ * * , ( ) ] link:\doibase 10.1016/j.jmmm.2006.10.304 [ * * , ( ) ] link:\doibase 10.1088/0034 - 4885/60/5/001 [ * * , ( ) ] link:\doibase 10.1103/physrevb.79.014410 [ * * , ( ) ] link:\doibase 10.1103/physrevb.55.12318 [ * * , ( ) ] link:\doibase 10.1103/physrevb.66.100403 [ * * , ( ) ] link:\doibase 10.1088/0953 - 8984/1/10/007 [ * * , ( ) ] link:\doibase 10.1103/physrevb.56.11678 [ * * , ( ) ] link:\doibase 10.1515/zkri-2014 - 1737 [ * * , ( ) ]
|
we study electronic and magnetic properties of the quasi - one - dimensional spin-@xmath0 magnet ba@xmath1cu@xmath1sc@xmath2o@xmath3 with a distinct orthogonal connectivity of cuo@xmath4 plaquettes .
an effective low - energy model taking into account spin - orbit coupling was constructed by means of first - principles calculations . on this basis a complete microscopic magnetic model of ba@xmath1cu@xmath1sc@xmath2o@xmath3 , including symmetric and antisymmetric anisotropic exchange interactions , is derived .
the anisotropic exchanges are obtained from a distinct first - principles numerical scheme combining , on one hand , the local density approximation taking into account spin - orbit coupling , and , on the other hand , projection procedure along with the microscopic theory by toru moriya .
the resulting tensors of the symmetric anisotropy favor collinear magnetic order along the structural chains with the leading ferromagnetic coupling @xmath5mev .
the interchain interactions @xmath6mev and @xmath7mev are antiferromagnetic .
quantum monte carlo simulations demonstrated that the proposed model reproduces the experimental neel temperature , magnetization and magnetic susceptibility data .
the modeling of neutron diffraction data reveals an important role of the covalent cu
o bonding in ba@xmath1cu@xmath1sc@xmath2o@xmath3 .
|
transition - metal dichalcogenides , similar to graphite , are layered materials with weak interlayer van der waals interaction .
these materials can be exfoliated to single to few - layer samples as the fabrication of graphene.@xcite among these samples , monolayer molybdenum disulphide ( mos@xmath0 ) has received much attention due to its distinctive properties.@xcite it has a direct gap at the inequivalent k and k@xmath1 points,@xcite which makes it attractive as a two - dimensional channel material in field - effect transistors .
very recently , the field - effect transistor devices based on monolayer mos@xmath0 have been realized in the experiments with high on - off ratio.@xcite in addition , space inversion symmetry is broken in monolayer mos@xmath0 since it is a two - dimensional hexagonal lattice consisted of two different sublattices , i.e. , mo and s atoms .
the absence of the space inversion symmetry results in a valley - dependent optical selection rule for interband transitions , which allows the realization of the valley polarization by optical pumping with circularly polarized light.@xcite space inversion symmetry breaking can also induce spin splitting of both the conduction and valence bands where the one of the valence band is much larger than that of the conduction band.@xcite this spin splitting is essential for spin physics and spintronic applications .
all these intriguing features make monolayer mos@xmath0 of particular interest . as spin relaxation is crucial for possible realistic spintronic applications ,
a thorough understanding of the spin relaxation time ( srt ) in monolayer mos@xmath0 is essential .
very recently , ochoa and roldn@xcite investigated the intravalley spin - orbit mediated spin relaxation in monolayer mos@xmath0 . for electrons , they calculated the out - of - plane spin relaxation due to the extrinsic rashba spin - orbit coupling ( soc ) induced by an external out - of - plane electric field and also the in - plane one due to the intrinsic soc from the contribution of the valence band due to the space inversion symmetry .
however , according to the latest report by kormnyos _
et al._,@xcite this intrinsic soc from the contribution of the valence band is weak since the splitting of the valence band is much smaller than the typical band gap .
in contrast , the intrinsic soc from the conduction band , which manifests itself as a zeeman - like term with opposite effective magnetic fields in the two valleys , is absent in their calculation.@xcite the zeeman - like term can give rise to an intervalley spin relaxation channel in the presence of intervalley electron - phonon scattering , which has been shown of significant importance to the spin relaxation in both rippled single - layer graphene@xcite and also bilayer graphene.@xcite moreover , only the disorder is included in a phenomenological manner in their calculation . the electron - electron coulomb and
electron - phonon interaction , which have been demonstrated to play an important role in spin relaxation in semiconductors@xcite and also graphene at high temperature,@xcite are absent . in the present work , with the electron - electron coulomb , ( both the intra- and inter - valley ) electron - phonon and the electron - impurity scatterings explicitly included , we study the electron spin relaxation due to the dyakonov - perel@xcite mechanism in monolayer mos@xmath0 in the absence of external electric field by the kinetic spin block equation ( ksbe ) approach.@xcite according to the latest report by kormnyos _
et al._,@xcite the effective magnetic field of the intrinsic soc of the conduction band is given by @xmath2 with the @xmath3-axis being perpendicular to the monolayer mos@xmath0 plane .
here , @xmath4 and @xmath5 represent the strength of the soc and k(k@xmath1 ) valleys , respectively.@xcite it is noted that the contribution of the hybridization by the intrinsic soc of the valence band to the energy bands is neglected since the energy scale of this spin splitting is much smaller than the typical band separation.@xcite this effective magnetic field is momentum independent , indicating an absence of the intravalley inhomogeneous broadening@xcite for in - plane spins .
therefore , the intravalley spin relaxation process contributed by the electron - electron coulomb , intravalley electron - phonon and electron - impurity scatterings are irrelevant to the in - plane spin relaxation . here
, only the intervalley process contributes to the in - plane spin relaxation .
the underlying physics is the same as the case of bilayer graphene@xcite and also rippled single - layer graphene.@xcite this intervalley spin relaxation channel suppresses the in - plane srt effectively at high temperature .
we also find that this system is always in the weak intervalley scattering limit , which determines that the in - plane spin relaxation time is proportional to the intervalley momentum scattering time . as a result ,
a monotonic decrease of the in - plane srt with increasing temperature is shown .
we also find that the in - plane srt decreases with the increase of the electron density .
moreover , we show a decrease of the in - plane srt with increasing initial spin polarization at low temperature , which is very different from the previous studies in both semiconductors@xcite and single - layer graphene@xcite but similar to the case of bilayer graphene.@xcite
with the effective mass approximation , the hamiltonian of the conduction band near the k(k@xmath1 ) points in monolayer mos@xmath0 can be described by @xmath6 according to the latest report by kormnyos _
et al._.@xcite here , @xmath7 with @xmath8 and @xmath9 being the in - plane momentum relative to the k(k@xmath1 ) points and the effective mass , respectively .
@xmath10 are the pauli matrices and @xmath11 is given in eq .
( [ soc ] ) .
we then construct the microscopic ksbes@xcite to investigate the intrinsic electron spin relaxation in monolayer mos@xmath0 .
the ksbes are given by@xcite @xmath12 in which @xmath13 stand for the density matrices of electrons with the diagonal terms @xmath14 representing the distribution functions and the off - diagonal ones @xmath15 describing the spin coherence .
the coherent terms @xmath16 are given in ref . .
@xmath17 are the scattering terms including the electron - electron coulomb ( @xmath18 ) , electron - impurity ( @xmath19 ) , intravalley electron - acoustic phonon , electron - optical phonon , and especially the intervalley electron - phonon scattering including the electron - kta phonon , electron - kla phonon , electron - kto phonon and electron - klo phonon scatterings . here , kta , kla , kto and klo correspond to the transverse acoustic , longitudinal acoustic , transverse optical and longitudinal optical phonon modes at k point , respectively.@xcite the detailed expressions of the above scattering terms can be found in ref .
( note that the form factor in the electron - impurity and electron - electron coulomb scatterings in ref . is absent here ) .
the scattering matrix elements @xmath20 being the impurity charge number ( assumed to be 1 in our calculation ) .
the quantity @xmath21 stands for the screening under the random phase approximation.@xcite @xmath22 is the two - dimensional bare coulomb potential with @xmath23 being the relative static dielectric constant.@xcite the electron - phonon scattering matrix elements are explicitly given by the latest reports by kaasbjerg _
et al._.@xcite specifically , we lay out the matrix elements of the electron - kta ( @xmath24 ) and -klo ( @xmath25 ) phonon scatterings , which play a dominant role in the in - plane spin relaxation as will be shown later . @xmath26 where @xmath27 is the area of the sample ; @xmath28 is the mass density of the monolayer mos@xmath0 ; @xmath29 and @xmath30 are the energies of kta and klo phonon modes , respectively ; @xmath31 ( @xmath32 ) is the first- ( zeroth- ) order deformation potential corresponding to the electron - kta ( -klo ) phonon scattering ; @xmath33 .
it is noted that the screening by the carriers is included in the electron - phonon scattering in their works with the random phase approximation for acoustic phonon and debye - hckel one in the nondegenerate limit for other phonon modes.@xcite it is further noted that they pointed out that the screening can be neglected for the intervalley electron - phonon scattering due to the large wave vectors of phonons.@xcite therefore , in the present work , we only include the screening for the intravalley electron - phonon scattering with the random phase approximation.@xcite
in the literature , there still remain controversies over the band structure of monolayer mos@xmath0,@xcite where different energy gaps , effective masses and the spin splittings are given . here , we take the effective mass and the strength of the soc being @xmath34@xcite and @xmath35 mev,@xcite respectively .
@xmath36 stands for the free electron mass .
the relative static dielectric constant is chosen to be @xmath37.@xcite the mass density @xmath38g/@xmath39 ; the kta and klo phonon energies are @xmath40mev and @xmath41mev , respectively ; the deformation potentials for the kta and klo phonons are @xmath42ev and @xmath43ev / cm , separately.@xcite with these parameters , our results are obtained by numerically solving the ksbes [ eq . ( [ ksbe])].@xcite the initial spin polarization
is set to be @xmath44 % and the spin - polarization direction is chosen in the monolayer mos@xmath0 plane unless otherwise specified . as reported , in contrast to the out - of - plane spin orientation,@xcite
the in - plane one is difficult to generate optically due to the large out - of - plane spin splitting of the valence band .
however , with an in - plane magnetic field , the in - plane component of the spin polarization can be obtained from the out - of - plane spin orientation.@xcite in addition , the in - plane spins can also be realized electrically by ferromagnetic contacts.@xcite the electrical method has been widely used in spin transport measurements in semiconductors@xcite and also graphene.@xcite we first study the temperature dependence of the spin relaxation .
the in - plane srt @xmath45 as function of temperature @xmath46 is plotted in fig .
[ fig1](a ) .
it is seen that the srt decreases monotonically with the increase of the temperature . to understand this behavior
, we calculate the temperature dependence of the srt with only the electron - electron coulomb , electron - impurity , intravalley electron - acoustic phonon , electron - optical phonon , or intervalley electron - phonon scattering included , separately .
we find that the contributions of the electron - electron coulomb , electron - impurity , intravalley electron - phonon scatterings , which only influence the intravalley spin relaxation channel , are negligible due to the absence of a momentum - dependent effective magnetic field of the soc [ eq . ( [ soc])].@xcite therefore , the srt is only determined by the intervalley spin relaxation channel.@xcite with the zeeman - like term being @xmath47 [ eq . ( [ soc ] ) ] , the in - plane srt is given by @xmath48 according to the report by zhang _
et al._@xcite based on the elastic scattering approximation . here
, @xmath49 represents the intervalley electron - phonon scattering time .
it is noted that the intervalley electron - phonon scattering is always in the weak scattering limit for the electron density upto @xmath50@xmath51 , therefore @xmath52 throughout our work . as a result ,
the srt decreases with the enhancement of the intervalley electron - phonon scattering as the temperature increases .
in addition , we also show a comparison of contributions from two leading intervalley electron - phonon scatterings , i.e. , electron - kta and electron - klo ones in fig
. [ fig1](a ) .
we find that the electron - kta phonon scattering plays a more important role in spin relaxation at low temperature whereas the electron - klo phonon scattering becomes more important at high temperature .
this can be understood from the contribution of the electron - phonon scattering matrix element together with phonon number .
it is noted that the electron - kta phonon scattering matrix element is smaller than the electron - klo phonon one whereas the kta phonon mode has a larger phonon number due to a smaller phonon energy.@xcite at low temperature , the phonon number of kta mode is much larger than that of klo mode , which makes the electron - kta phonon scattering stronger and hence more important to the spin relaxation .
however , at high temperature , the phonon numbers of two modes become comparable .
then the electron - klo phonon scattering plays a more important role in spin relaxation due to the larger scattering matrix element .
it is noted that the contributions of the electron - kla and -kto phonon scatterings to the spin relaxation are marginal .
( @xmath53 ) and that calculated with only the electron - kta phonon ( @xmath54 ) or electron - klo phonon ( @xmath55 ) scattering included as function of temperature @xmath46 with the electron density @xmath56@xmath51 ; ( b ) @xmath45 as a function of @xmath57 at @xmath58k ; ( c ) total in - plane srt @xmath45 @xmath59 ( @xmath60 ) and that calculated without the coulomb hf term @xmath61 ( @xmath53 ) as function of the initial spin polarization @xmath62 at @xmath58k with @xmath63@xmath51.,title="fig:",width=204 ] ( @xmath53 ) and that calculated with only the electron - kta phonon ( @xmath54 ) or electron - klo phonon ( @xmath55 ) scattering included as function of temperature @xmath46 with the electron density @xmath56@xmath51 ; ( b ) @xmath45 as a function of @xmath57 at @xmath58k ; ( c ) total in - plane srt @xmath45 @xmath59 ( @xmath60 ) and that calculated without the coulomb hf term @xmath61 ( @xmath53 ) as function of the initial spin polarization @xmath62 at @xmath58k with @xmath63@xmath51.,title="fig:",width=204 ] ( @xmath53 ) and that calculated with only the electron - kta phonon ( @xmath54 ) or electron - klo phonon ( @xmath55 ) scattering included as function of temperature @xmath46 with the electron density @xmath56@xmath51 ; ( b ) @xmath45 as a function of @xmath57 at @xmath58k ; ( c ) total in - plane srt @xmath45 @xmath59 ( @xmath60 ) and that calculated without the coulomb hf term @xmath61 ( @xmath53 ) as function of the initial spin polarization @xmath62 at @xmath58k with @xmath63@xmath51.,title="fig:",width=204 ] then we turn to investigate the electron - density dependence of the in - plane srt and show the results at @xmath58k in fig .
[ fig1](b ) .
we find that the srt decreases rapidly with increasing electron density at @xmath64k but decreases mildly at @xmath65k .
we first focus on the case at @xmath64k .
the srt is dominated by the electron - kta and -klo phonon scatterings as pointed out in the temperature dependence of the spin relaxation . the electron - klo phonon scattering , which has a momentum - independent matrix element [ see eq .
( [ klo ] ) ] , is insensitive to the electron density .
however , the electron - kta phonon scattering is enhanced with increasing electron density as its matrix element [ see eq .
( [ kta ] ) ] increases with the increase of the momentum .
therefore , when the electron density increases , the enhancement of the electron - kta phonon scattering leads to the decrease of the srt ( @xmath52 ) .
as for the case at @xmath66k , the increase of the matrix element of the electron - kta phonon scattering becomes insensitive .
in addition , the electron - klo phonon scattering plays a more important role in spin relaxation than that in the case at @xmath64k .
both lead to a mild decrease of the srt . in the previous investigations in both semiconductor systems@xcite and single - layer graphene,@xcite
a significant increase of the srt is shown with the increase of the initial spin polarization , which originates from the contribution of the coulomb hartree - fock ( hf ) term .
however , in the case of bilayer graphene , the srt decreases rapidly with increasing initial spin polarization at low temperature whereas shows a mild increase at high temperature.@xcite in this case , the srt is determined by the intervalley electron - phonon scattering . here
, we also investigate the initial spin polarization dependence of spin relaxation . in fig .
[ fig1](c ) , we plot the in - plane srt as function of the initial spin polarization @xmath62 at @xmath58k .
we find that the srt shows a decrease with increasing @xmath62 at @xmath64k whereas becomes insensitive to @xmath62 at @xmath66k .
we first focus on the case of @xmath64k .
the decrease of the srt at @xmath64k is very different from the previous studies in both semiconductors@xcite and single - layer graphene@xcite but similar to the case of bilayer graphene at low temperature.@xcite the underlying physics is understood as follows . the srt is also determined by the intervalley electron - phonon scattering whereas the contribution of the coulomb hf term is negligible to the spin relaxation by comparing the calculation with and without the coulomb hf term shown in fig .
[ fig1](c ) .
the intervalley electron - phonon scattering is in the weak intervalley scattering limit , i.e. , the srt @xmath52 [ eq . ( [ formulazhang ] ) ] . as a result , the srt decreases with the increase of the initial spin polarization due to the enhancement of the intervalley electron - phonon scattering.@xcite as for the case of @xmath66k , the srt is also determined by @xmath52 [ eq . ( [ formulazhang ] ) ] .
the insensitive initial spin polarization dependence of @xmath67 originates from that of @xmath49 at @xmath66k.@xcite
as mentioned previously , the effective magnetic field [ see eq .
( [ soc ] ) ] is momentum independent , which leads to the absence of the intravalley relaxation channel for in - plane spins .
this effective magnetic field can become momentum dependent when the contribution of the hybridization by the intrinsic soc of the valence band to the energy bands is taken into account.@xcite this intravalley inhomogeneous broadening together with the intravalley scattering opens an intravalley spin relaxation channel .
the intravalley process competes with the intervalley one discussed here and the competition between these two processes needs further investigations.@xcite in addition , in real systems , the effective magnetic field can be sensitive to the random local strains@xcite due to the symmetry breaking , which also results in an intravalley inhomogeneous broadening and hence the intravalley spin relaxation process .
however , this is out of the scope of the present work since the effect of the random local strains on the spin - orbit coupling is still unclear in monolayer mos@xmath0 . in summary
, we have investigated the intrinsic electron spin relaxation due to the dyakonov - perel mechanism in monolayer mos@xmath0 .
the effective magnetic field of the intrinsic soc is momentum independent , which supplies a zeeman - like term with opposite effective magnetic fields perpendicular to the mononlayer mos@xmath0 plane in the two valleys .
since it is independent on momentum , the intravalley spin relaxation channel is absent for the in - plane spins .
however , the zeeman - like term , together with the intervalley electron - phonon scattering , gives rise to an intervalley relaxation channel for in - plane spin polarization , which is similar to the case in rippled single - layer graphene and also bilayer graphene .
this effect has not been yet reported in the literature on monolayer mos@xmath0 .
the intervalley spin relaxation channel can markedly suppress the in - plane srt at high temperature .
in addition to the intervalley electron - phonon scattering , the electron - impurity , electron - electron coulomb and intravalley electron - phonon scatterings are also included to calculate the in - plane srt .
however , their contribution to in - plane spin relaxation is negligible due to the absence of intravalley relaxation process .
moreover , we find that the intervalley scattering is always in the weak scattering limit in this material .
therefore , the srt is always proportional to the intervalley momentum scattering time .
this results in a monotonic decrease of the in - plane srt with the increase of temperature .
a decrease of the in - plane srt is also observed in the electron density dependence .
in addition , we find that the in - plane srt decreases with the increase of the initial spin polarization at low temperature , which is very different from the previous studies in both semiconductors and single - layer graphene but similar to the case of bilayer graphene .
this work was supported by the national natural science foundation of china under grant no .
11334014 , the national basic research program of china under grant no .
2012cb922002 and the strategic priority research program of the chinese academy of sciences under grant no . xdb01000000 .
|
intrinsic electron spin relaxation due to the dyakonov - perel mechanism is studied in monolayer molybdenum disulphide .
an intervalley in - plane spin relaxation channel is revealed due to the opposite effective magnetic fields perpendicular to the monolayer molybdenum disulphide plane in the two valleys together with the intervalley electron - phonon scattering .
the intervalley electron - phonon scattering is always in the weak scattering limit , which leads to a rapid decrease of the in - plane spin relaxation time with increasing temperature .
a decrease of the in - plane spin relaxation time with the increase of the electron density is also shown .
|
various effects of flooding can drastically alter an ecosystem . some of the factors that could potentially harm trees include alteration of soil factors , disturbance of plant life , and degeneration of habitat . during a flood , soil is altered by erosion , deoxygenation , and reconfiguration ( kozlowski 1997 ; lake 2011 ) .
erosion is caused by the continuous circular and turbulent motion of the water .
this motion carries particles of the soil away , resulting in a loss of topsoil , loss of nutrients , and an exposure of plant roots ; all three of these factors severely harm plants ( shafroth et al .
the topsoil is reconfigured with the finer particles that were carried with the flood , such as sand and silt , which may to compact into finer arrangements .
this compacted soil restricts the ability to hold gases such as oxygen , causing deoxygenation of the soil ( parrett 1964 ) .
flooding disturbs not only the abiotic factors of the environment , but also the biotic factors . plants suffocate and decompose due to the flood waters ( parolin & wittmann 2010 ) .
this can result in stunting , damaging , or even killing of trees ( joly & crawford 1982 ) .
the warm , stagnant water is a suitable habitat for several bacteria and fungi .
this can cause tree rot as the bacteria and fungi in the water inhibit the trees natural ability to carry nutrients and water ( coder 1994 ) . because the part of a tree is submerged
, the tree is likely to undergo suffocation if it is submerged for a long period of time .
the suffocation of trees is caused by the lack of oxygen in the stagnant flood waters .
if there is a faster flowing current , the water can also damage a tree s structure by breaking off branches or uprooting it altogether .
considering both abiotic and biotic factors , flooding substantially distorts the original environment .
damage to native plant life during the flood and the change in soil configuration after the flood may provide a more suitable environment for foreign species ( yanosky 1982 ) .
since the postdiluvian soil determines the species of plants present , a flood can restructure an ecosystem ( bratkovich 1993 ) .
minot , north dakota , is positioned in the souris river valley .
the souris river originates in canada , flows south into the united states at north dakota , passes through minot , and then turns north again to re - enter canada .
flooding of varying severity is common along the souris river . despite numerous previous floods ,
none were as large as the flood of 2011 .
record water levels from the 1969 flood were surpassed as water rose to 457.8 m above sea level .
this record flood event submerged houses and crested bridges .
warm , stagnant , and hypoxic waters from summer rains covered the valley for approximately 23 days in june and july ( `` timeline ... '' ) .
the river banks are presently wider and contain more silt than the antediluvian banks ; this was caused by erosion of the banks .
the flood has also affected the plant life present in the flood plain .
vast areas of the minot flood plain were drastically over - saturated during the flood , causing a degeneration of suitable habitats for most indigenous plant species ( `` timeline ... '' ) . studying tree species in minot has allowed for better understanding of the extent of damage suffered by trees during floods . by determining which trees are best adapted to survive in flooding situations , informed decisions can be made when planting trees in areas prone to flooding .
this knowledge will allow for the least sustained damage of habitats after future floods . here , we analyze data collected regarding the impact of a flood event on tree damage .
we hypothesize that tree height , genus , and maximum water level are all determining factors in the severity of damage sustained by a tree in a flood event .
we used the minot flood plain for research .
we sampled trees across the minot flood plain in september 2011 and in oak park in july 2014 with tools such as clinometers , logger s tape , and a gps device . for each tree we recorded height , water level , identity , and damage .
damage was visually accessed on a scale of 0 to 5 : 0 = no damage , 1 @xmath0 25% of the tree was damaged , 2 = 2550% of the tree was damaged , 3 = 5075% of the tree was damaged , 4 = 7590% of the tree was damaged , and 5 = 100% of the tree was damaged .
we measured tree heights with a clinometer and recorded to nearest meter .
maximum water level was measured using the still visible line of water on objects surrounding the tree , such as houses or fences .
we identified ( herman & chaput 2003 ) to the level of genera and measured 288 trees . the second part of the research explored the relationship between the maximum water level and tree damage .
we made a virtual transect along the 8th street ( which comes almost perpendicular from the river to the undamaged part of city ) and use apple trees ( _ pyrus _ ) as a standard indicator of damage .
we recorded the damage and distance from the river in meters of _ pyrus _ trees along this route . in the oak park section ,
the exact gps coordinates of trees were also recorded .
this measurement will allow for accurate and consistent reevaluation over time .
the damage rating has been recorded annually .
we used the r statistical environment ( r core team 2014 ) for all calculations and plots .
data categories were analyzed in various combinations in order to thoroughly evaluate relationships ( dalgaard , 2008 ) .
we analyzed the relationship between height and genera , water level and genera , tree height and water level , tree height and damage , water level and damage , genera and damage , conifers and damage , angiosperms and damage , and damage and time .
relationships between the data were all determined after inputting the data .
we analyzed the data through these various combinations in order to fully understand the relationships between the factors .
all data files and r script are available in the open data repository at http://ashipunov.info/shipunov/open/ecol_impact_flooding.zip
288 studied trees belonged to 12 different genera ; the majority being _ fraxinus _ and _ ulmus_. damages recorded were mostly below 3 ; less than 40% trees were damaged > 50% .
we measured interactions between the data recorded .
both height of tree and maximum water level were significantly correlated with the taxonomy ( kruskall - wallis tests p - values @xmath1 in both cases ) .
tree height and water level were also weakly but significantly correlated ( spearman @xmath2 , p @xmath1 ) .
consequently , on next steps of the data analysis we focused on the interaction between factors . the linear models of damage included two continuous ( tree height and water level ) and one categorical variable ( tree genus ) , and were different with and without interactions . with interactions , only one term ( _ quercus _ genus ) was significant whereas the model without interactions had multiple significant terms , namely tree height , water level , and _ prunus _ , _ pyrus _ and _ quercus _ genera ( all p - values < 0.05 ) . _ prunus _ and _ pyrus _ sustained damage more than other trees ( median damage 3 while the median damage for the all data was 2 ) , but _
quercus _ was damaged on the median level .
both linear models were significant ( fisher tests p - values @xmath1 ) , but the adjusted r@xmath3 in both cases was low ( 0.167 and 0.111 , respectively ) . the second model was more optimal ( aic @xmath4 vs. aic @xmath5 for the first model with interactions ) . attempts to remove any terms in order to simplify the second model resulted in an increased aic .
the damage was weakly but significantly negatively correlated with the tree height ( spearman @xmath6 , p - value @xmath1 ) but not with water level ( spearman @xmath7 , p - value @xmath8 ) .
post hoc pairwise comparisons of damage level between different tree genera did not reveal any significant patterns ( nemenyi test p - values @xmath9 ) , only damages to _ pyrus _ and _ populus _ were significantly different ( nemenyi test p - value @xmath10 ) .
various contrasts were applied to our set of tree species .
while native and non - native trees were not significantly different in the level of damage ( wilcoxon test p - value @xmath11 ) , conifers were damaged significantly more than angiosperm trees ( wilcoxon test p - value @xmath12 ) . however , without _ juniperus _
( i.e. , only for pine family , pinaceae trees ) this significant pattern disappeared ( wilcoxon test p - value @xmath13 ) .
rosaceae trees ( _ pyrus _ and _ prunus _ ) also suffered damage more than others ( wilcoxon test p - value @xmath14 ) but this could have been influenced by their significantly smaller sizes ( wilcoxon test p - value @xmath1 ) . to understand how damage is related with water level , we used transect data less influenced by interactions since it focused on only one genus ( _ pyrus _ ) and location . the linear model for the damage vs. maximum level of water on
the transect returned all terms significant , adjusted r@xmath3 @xmath15 and fisher test p - value @xmath1 . while there was a decreasing trend of the tree damage in oak park ( from 3.0 in 2011 to 2.0 two years later ) , this pattern was not significant ( friedman test p - value @xmath16 ) .
the patterns observed suggest several conclusions .
first of all , the assumption of a greater tolerance to flooding of native species than nonnative species is not supported by the data .
others have observed a relationship between indigenous species and damage resistance in response to locally common disasters ( zamora - arroyo et al .
2001 ) , but this was not supported by our flood data .
this is probably because at least some of the foreign species originated from areas with similar flooding occurrences as the minot flood plain
. a significant inverse relationship between tree height and damage sustained was found . while tree height contributes greatly to the amount of damage trees
sustain after a flood , the data also shows the separate relationships between the taxonomy and damage .
the result that certain genera show greater resistance to flooding than other genera is consistent with previous research ( parolin & wittmann 2010 ; kozlowski et al . , 2015 ) .
conifers ( pinopsida ) and _ pyrus _ and _ prunus _ ( both of rosaceae family ) sustained significantly more damage than other genera .
these results allow for educated decisions in the future when planting trees inside a flood zone .
the factors measured were highly interactive in determining flood damage .
water level , tree height , and genus all had significant effects on tree damage ; the removal of any of these terms from the model decreased optimality .
while it appears that a combination of all factors determines damage , tree height seems to exhibit the greatest influence .
the study of damage to oak park trees over time is still in its early stages .
as it stands , the data suggest promising regeneration rates for most of the trees involved .
continued research will be conducted on these trees to gain further insight regarding tree regeneration .
drawing upon similar research , we predict that native trees will show a greater rate of regeneration than nonnative trees ( zamora - arroyo et al .
thank you to joe super ( minot high school ) for providing equipment and motivation .
many thanks to the students of minot state university biology 154 class for their contributions in data collection .
we are grateful to reviewers for their valuable suggestions and corrections .
bratkovich , s , l. burban , s. katovich , c. locey , j. pokorny , and r. wiest . 1993 .
flooding and its effect on trees .
newtown square , pa : u.s . dept . of agriculture , forest service , northern area state & private forestry .
coder , k. d. 1994 .
flood damage to trees .
university of georgia school of forest resources extension , publication for 94 - 061 .
accessed 22 november 2014 .
dalgaard , p. 2008 .
introductory statistics with r. springer science & business media , new york .
parolin , p. , and f. wittmann .
struggle in the flood : tree responses to flooding stress in four tropical floodplain systems .
aob plants 2010 : plq003 .
herman , d. e. , and l. j. chaput .
trees and shrubs of north dakota .
ndsu , fargo .
joly , c. a. , and r. m. m crawford .
variation in tolerance and metabolic responses to flooding in some tropical trees .
journal of experimental botany .
33:799809 .
kozlowski , g. , stoffel , m. , betrisey , s. , cardinaux , l. , and mota , m. 2015 .
hydrophobia of gymnosperms : myth or reality ? a global analysis . ecohydrology .
8:105112 .
kozlowski , t. t. 1997 .
responses of woody plants to flooding and salinity .
tree physiology monograph .
1:129 .
lake , p. s. 2011 .
drought and aquatic ecosystems : effects and responses .
chichester , west sussex . : wiley blackwell .
parrett , a. 1964 .
montana s worst natural disaster .
montana : the magazine of western history 2:20 .
r core team . 2014 .
r ; a language and environment for statistical computing .
r foundation for statistical computing , vienna , austria .
shafroth , p. b. , j .c .
stromberg , and d. t. patten . 2000 .
woody riparian vegetation response to different alluvial water table regimes .
western north american naturalist , 60:6676 .
timeline of minot s 2011 flood .
2011 august 07 .
minot ( nd ) : minot daily news .
http://www.minotdailynews.com/page/content.detail/id/ 557475/timeline - of - minot - s-2011-flood.html?nav=5010[http://www.minotdailynews.com / page / content.detail / id/ 557475/timeline - of - minot - s-2011-flood.html?nav=5010 ] .
accessed 16 march 2015 .
yanosky , t. m. 1982 .
effects of flooding upon woody vegetation along parts of the potomac river flood plain .
geological survey professional paper 1206 , distr .
branch usgs .
zamora - arroyo , f. , p. nagler , m. briggs , d. radtke , h. rodriquez , j. garcia , c. valdes , a. huete , and e. glenn .
regeneration of native trees in response to flood releases from the united states into the delta of the colorado river , mexico .
journal of arid environments 49:4964
|
the objective of this research was to identify factors affecting tree damage in the historical minot flood of 2011 .
we hypothesized that tree height , identity , origin , and maximum water height affect in the severity of damage sustained by a tree in a flood event .
all these factors were significant but highly interactive .
the results from this research can influence planting practices in valleys and other flood prone areas to mitigate future damage .
|
numerical approximations of the perfectly matched layer ( pml ) @xcite for maxwell s equations in first order form often exhibit a late time linear or exponential energy growth , see for example @xcite .
this unwanted growth can spread into the computational domain and ruin the accuracy of a numerical simulation .
growth is often seen long after a wave has exited the domain or when a wave has been trapped in the domain for a longtime .
the later situation can occur , for example , in the simulation of an electromagnetic waveguide . this has generated substantial interests .
one explanation for this behavior of the pml was offered in @xcite .
it is shown that the split
field pml is weakly hyperbolic , and numerical approximations can lead to perturbations that excite explosive modes in the pml . because of the result in @xcite , several unsplit ( strongly hyperbolic ) pmls for maxwell s equations were developed @xcite .
however , series of numerical experiments presented in the literature , see for instance @xcite , using standard finite difference schemes and @xcite for finite element methods , suggest that these unsplit pmls also suffer from the same late time exponential energy growth as the split field pml .
it was later shown in @xcite that the source of the late time exponential growth was characterized by the multiplicity of eigenvalues and the degeneracy of eigenvectors of the lower order undifferentiated terms . in @xcite , however , it was shown , for a given computational setup , that the growth generated in the pml , if any , may not propagate back into the computational domain .
it is also important to note that a partial remedy for this problem in odd ( 1 and 3 ) spatial dimensions has been proposed in @xcite , using a lacunae based time integration .
nonetheless , a recent analysis of a pml for maxwell s equations in @xcite shows that the corresponding initial value problem ( ivp ) is asymptotically stable . in this study
, we argue that many of the late time energy growth of the pml for maxwell s equations observed in computations are caused by inadequate numerical boundary treatments in the pml .
we denote that the well posedness and temporal stability analysis of the pml has been a topic of several works , see for instance @xcite and many others . in @xcite , perfect matching , well - posedness and temporal stability for pmls as well as some other classes of layers are discussed for general systems .
unfortunately , there is no guarantee that all solutions decay with time . in @xcite
however , the geometric stability condition was introduce to characterize the temporal stability of ivp for the pml .
if this condition is not satisfied , then there are modes of high spatial frequencies with temporally growing amplitudes . in this paper
, we consider a two space dimensional model problem satisfying the geometric stability condition but augmented with boundary conditions in the tangential and normal directions .
our primary objective in this paper is the analysis and development of high order accurate and provably stable numerical approximations of the corresponding initial boundary value problem ( ibvp ) .
it is a well known fact that the theory and numerical techniques to solve an ibvp are more elaborate and complicated than that of a corresponding ivp . for a
well posed partial differential equation which is stable in the absence of boundaries can support unstable solutions , or become ill posed , when boundaries are introduced .
in fact , the analysis of the ivp is a necessary ( first ) step towards the analysis of the ibvp @xcite . for symmetric hyperbolic systems such as the maxwell s equations and certain boundary conditions , using the energy method , it is possible to obtain a uniform bound on the l@xmath0norm of the solutions ( electric and magnetic fields ) in terms of the l@xmath0norm of the initial data and a boundary norm of the boundary data @xcite . in numerical approximations of ibvps ,
most of the difficulties arise from the boundaries .
similarly , a numerical method which is stable in the absence of boundaries can support growth when boundary conditions are imposed . in the discrete
setting , it is possible though to design stable numerical boundary procedures by mimicking the continuous estimates @xcite .
effective boundary treatments are also important when pmls are used in computations . if there are physical boundaries , for instance in waveguides @xcite , the corresponding boundary conditions must be accurately extended into the pml . in practice
, the pml must be solved on a bounded computational domain .
a stable and accurate pml boundary closure becomes essential , since it enables efficient numerical computations .
there is an ongoing discussion though on how to effectively terminate the pml .
it is also important to point out that a numerical method which is stable in the absence of the pml can support growth when the pml is included .
a major difficulty is that in general the pml ( for symmetric systems ) is asymmetric .
therefore , deriving energy estimates for the pml that are useful in designing stable and accurate numerical boundary procedures can be very challenging .
linear and/or exponential temporal growth are frequently seen in numerical simulations of pmls @xcite , particularly if higher order methods are considered .
often , artificial numerical dissipations are added to the scheme in order to eliminate these unwanted temporal growth .
artificial numerical dissipations certainly helps in many ways , but can also introduce some interesting , and unfortunately undesirable , effects . in standard computational electromagnetic codes , discrete stability of the numerical method is usually investigated in the absence of the pml . once the interior scheme is confirmed stable , the pml is then included by discretizing the accompanying auxiliary differential equations and appending the auxiliary functions accordingly , without further discrete stability analysis . as in @xcite , any eventual growth seen when the absorption function is present
is then attributed to the pml .
we emphasize that a more complete analysis of discrete stability must include all the equations and variables in the pml . in
bounded and semi bounded domains the stability of the pml has been analyzed for certain classes of problems and boundary conditions @xcite using normal mode analysis .
these results @xcite are mostly for second order systems . in the continuous
setting , it is possible to extend some of these theoretical results to the pml for first order systems .
however , numerical treatments for first order systems are essentially different from numerical methods for second order systems .
for instance , standard boundary conditions for first order hyperbolic systems can only contain a linear combination of the unknown variables , while boundary conditions for second order systems often involve combinations of the unknown variables and their spatial and temporal derivatives . therefore , for first order systems different numerical techniques and , probably , new theories are required . in the present paper
, we consider unsplit pmls @xcite and the classical split
field pml @xcite , for maxwell s equations in first order form with boundary conditions set by an electromagnetic waveguide .
we are particularly interested in high order accurate and efficient methods for numerical simulation of maxwell s equations .
our first objective is to extend the continuous analysis for cauchy problems @xcite to the pml in semi bounded and bounded domains .
the first sets of results are the proofs of stability of the constant coefficient pml in a lower half plane and a left half plane .
the second results in the paper are the derivation of energy estimates for constant coefficients in the laplace space and for variable coefficients pml in the time domain .
we show that a stable numerical boundary procedure for the undamped problem can become unstable when the pml is introduced .
the second objective of the paper is to develop a high order accurate and provably stable numerical approximation for the pml in a bounded domain .
we use summation by parts ( sbp ) finite difference operators @xcite to approximate spatial derivatives .
boundary conditions are enforced weakly using a penalty technique , referred to as the simultaneous approximation term ( sat ) methodology @xcite . by mimicking the continuous energy estimates we construct accurate and stable boundary procedures .
the convergence of the numerical method follows .
finally , we present numerical experiments , using sbp finite difference operators in space with interior order of accuracy 2 , 4 and 6 , and the classical fourth order accurate runge kutta scheme in time .
the numerical examples illustrate high order accuracy and longtime stability of the pml .
note that in @xcite it was pointed out that the discrete pml for maxwell s equations in first order form can suffer severely from longtime exponential energy growth when spatial derivatives are discretized with high order accurate central finite difference operators and and the time derivative discretized using runge kutta schemes .
this indicates that an appropriate remedy is required .
the results obtained in this paper provide a cure for this problem without any additional cost .
we believe that the results obtained in this paper can be extended to other situations , not only for finite difference electromagnetic codes , but also for finite volume methods , finite element methods and spectral methods , and for more complicated systems such as the elastic wave equation in isotropic media and the linearized euler equation with vanishing mean flow .
the remainder of the paper is planned as follows . in the next section
, we present some preliminaries , introducing maxwell s equations in two space dimensions and a brief description of a discrete approximation of the maxwell s equations in a bounded domain using the sbp - sat methodology . in section 3
, we introduce the pmls and present some instructive numerical experiments , motivating the continuous analysis in section 4 and the discrete analysis in section 5 . in section 5 , we construct discrete approximations for the pml , derive discrete energy estimates and prove asymptotic stability .
more numerical experiments are presented in section 6 , verifying the analysis of previous sections .
a summary of the paper and suggestions for future work are offered in the last section .
here , we introduce the maxwell s equations in a two dimensional rectangular domain .
we end the section with a brief description of the sbp sat methodology for maxwell s equations .
consider the transverse magnetic @xmath1 case of the maxwell equation in a source free , homogeneous isotropic medium , @xmath2 here , @xmath3 are the magnetic fields and electric fields respectively . at the initial time
@xmath4 we set the smooth initial data @xmath5 consider now the two space dimensional rectangular domain @xmath6 \times [ -y_0 , y_0]$ ] , with @xmath7
. the length and width of the rectangle are @xmath8 and @xmath9 . at the edges of the rectangle , at @xmath10 and @xmath11 , we set the general boundary conditions , expressed as a relation between the incoming and outgoing characteristics @xmath12 @xmath13 here , the reflection coefficients @xmath14 @xmath15 with @xmath16 , are real constants .
note that the boundary condition , can support boundary waves : waves whose amplitudes are maximum on the boundaries , at @xmath10 , @xmath11 , but decay exponentially into the domain . in particular if @xmath17 , then the boundary conditions , model insulated electromagnetic walls . glancing waves , constant modes normal to the boundaries , can be supported .
if @xmath18 , then the boundary conditions , model perfect electric conductor ( pec ) boundary conditions . and if @xmath19 , then the boundary conditions , correspond to local one dimensional absorbing boundary conditions , which are also equivalent to setting the incoming characteristics ( riemann invariant ) at the boundaries to zero .
note that if @xmath20 then we have @xmath21 @xmath22 the special case of @xmath23 in , corresponds to pec boundary conditions @xmath24 @xmath25 let the standard @xmath26 inner product and the corresponding norm be defined by @xmath27 we also introduce the boundary norms @xmath28 , @xmath29 , defined by @xmath30 here , @xmath31 are the complex conjugates of @xmath32 and @xmath33 depend on one spatial variable @xmath34 or @xmath35 . using the energy method and integration by parts , it is straightforward to show that the maxwell s equation subject to the boundary conditions , , with @xmath36 , satisfies @xmath37 in particular if @xmath38 in , or @xmath18 ( we consider the pec condition ) in , then @xmath39 note that the @xmath26-norm of the solutions ( electric and magnetic fields ) are uniformly bounded in time by the @xmath26-norm of the initial data .
this subsection serves as a brief introduction to the sbp sat scheme for ibvps ( maxwell s equations in bounded domains ) .
the main idea is to approximate all spatial derivatives in using sbp finite difference operators .
the boundary conditions , are then enforced weakly using penalties .
however , penalty parameters must be chosen such that discrete energy estimates , analogous to the continuous estimates , , can be derived .
thus , proving asymptotic stability for the discretization .
more elaborate discussions on this topic can be found in @xcite , see also the recent review paper @xcite . to begin with ,
consider the uniform discretization of the unit interval with @xmath40 number of grid points and the spatial step , @xmath41 , @xmath42 let @xmath43 denote a differentiation matrix approximating the first derivative , @xmath44 on the grid .
the matrix @xmath43 is an sbp operator ( in one space dimension ) , see @xcite , if the following properties hold @xmath45 here , @xmath46 , @xmath47 pick out the left and right boundary values .
the matrix @xmath48 is symmetric and positive definite , and thus defines a discrete norm .
the matrix @xmath49 is almost skew - symmetric .
it is important to note that sbp operators can be constructed using galerkin finite / spectral element approaches @xcite .
the one space dimensional sbp operator @xmath43 can be extended to higher space dimensions using the kronecker product , @xmath50 , defined by [ def : section26 ] let a be an m - by - n matrix and b be a p - by - q matrix .
then @xmath51 , the kronecker product of a and b , is the ( mp)-by-(nq ) matrix + @xmath52{4}\\ \hdotsfor[4]{4}\\ a_{m1}b&a_{m2}b&\hdots&a_{mn}b \end{pmatrix}.\ ] ] the following properties hold for kronecker products 1 .
assume that the products ac and bd are well defined then + @xmath53 , 2 .
if a and b are invertible , then @xmath54 , 3 .
@xmath55 .
now consider the maxwell s equation in a two dimensional rectangular domain . to begin
, we discretize the spatial domain in the @xmath56- and @xmath57-directions using @xmath58 grid points with constant spatial steps @xmath59 , respectively .
a two dimensional scalar grid function , @xmath60 $ ] , is stacked as a vector of length @xmath61 , @xmath62 spatial derivatives are discretized using sbp operators , and the boundary conditions , and , are impose weakly using sat @xcite .
the sbp sat approximation of the maxwell s equation subject the boundary conditions , and , is [ eq : maxwell_waveguide_discrete ] @xmath63 here , the operators @xmath64 are identity matrices , and @xmath65 are the penalty parameters chosen by requiring stability .
note that the boundary conditions , are enforced weakly by the sat
terms , @xmath66 and @xmath67 , defined on the right hand side of equation .
this is opposed to strong enforcement of boundary conditions , where boundary conditions are explicitly imposed by injection . by construction
the sbp sat scheme is a consistent approximation of the ibvp , , and .
the critical aspect of the scheme lies in chosing penalties and ensuring numerical stability . since the ibvp , , , , does not support any temporal growth , it is imperative that penalties parameters must be chosen in a manner that does not allow any growth in time .
the penalty parameters , @xmath65 , are not necessarily unique
. often the choice of stable penalty parameters are guided by constructing discrete energy estimates which , as far as possible , mimic the continuous energy estimates , . in particular , penalty parameters are chosen such that the corresponding discrete boundary terms dissipate energy , at least , as fast as the continuous problem . however , from a practical point of view , for efficient explicit time integration , penalty parameters must also be chosen in order to avoid numerical stiffness .
note that if @xmath68 , then the ordinary differential equation ( ode ) is increasingly stiff . for more elaborate discussions and results on numerical boundary procedure
, please see @xcite .
note that the weak boundary treatment is often used for high order accurate discontinuous galerkin ( dg ) finite element methods @xcite and also for pseudo spectral methods @xcite .
we will now chose the penalty parameters @xmath65 such that the semi
discrete approximation is stable . to begin with
, we define the discrete scalar product and the corresponding discrete norm @xmath69 note that from the sbp property we have @xmath70 thus , using the sbp property we can rewrite the right hand side of to obtain [ eq : maxwell_waveguide_discrete_sbp ] @xmath71 by applying the energy method to , that is , we multiply , , from the left with @xmath72 , @xmath73 , @xmath74 , respectively , and add the transpose of the products , we obtain @xmath75 therefore , the discrete approximation is asymptotically stable if the boundary terms @xmath76 are nonnegative .
if @xmath77 and @xmath78 , then we can choose @xmath79 to obtain @xmath80 where @xmath81 are @xmath82 symmetric matrices and @xmath83 to ensure numerical stability the symmetric matrices @xmath84 defined above must have nonnegative eigenvalues .
the eigenvalues of the matrices @xmath84 are @xmath85 clearly , the eigenvalues are real and nonnegative if @xmath86 by comparing @xmath87 to the continuous analog in the right hand side of we can determine how fast energy is dissipated by the numerical boundary treatment . note that with @xmath88 , the boundary treatment dissipates energy as fast as the continuous problem , and we have @xmath89 thus , implying @xmath90 note that is completely analogous to .
note also that standard finite element and nodal finite volume methods for maxwell s equation are constructed to satisfy analogous discrete equations and energy estimates and .
if we choose @xmath91 @xmath92 , then the numerical boundary treatment will dissipate energy faster than the continuous problem . more generally , for any @xmath93 and @xmath94 ,
then we can choose @xmath95 to obtain @xmath96 note that @xmath97 and @xmath98 thus , the discrete solutions of are uniformly bounded in time by the initial data .
therefore , the discrete approximation with @xmath99 is also asymptotically stable .
the sbp operators used in the present study are high order accurate finite difference operators with diagonal norms @xmath48 , where the boundary stencils are @xmath100th order accurate and the interior accuracy is @xmath101 as will be seen in the next section , the pml will be placed in the @xmath56-direction .
the primary goal of this paper is to study the numerical difficulties , such as stability and stiffness , arising from the numerical boundary procedure when the pml is introduced .
we will also investigate the optimal penalty parameters for the pml .
the pml will be terminated by the characteristic boundary condition , with @xmath102 .
it is possible though to terminate the pml with the pec boundary condition , with @xmath103 .
however , the characteristic boundary condition at the external boundary have been shown to yield smaller spurious reflections , due to the modeling error , when combined with the pml than the pec backed pml @xcite . in order to closely make comparisons with the results published in the literature @xcite we will impose the characteristic boundary condition , with @xmath104 , in the @xmath57direction .
nevertheless , it is possible though to use the pec boundary condition , with @xmath105 , also in the @xmath57direction .
let us now consider the solutions of the maxwell s equations , in the truncated domain @xmath106 , where @xmath107 are artificial boundaries introduced to limit the computational domain .
in order to absorb outgoing waves at @xmath107 , we have introduced two layers , of width @xmath108 , having @xmath109 in which the pml equations are solved .
note that in the @xmath56-direction the boundary condition is and in the @xmath57-direction the boundary condition is .
this setup can model an infinitely long electromagnetic waveguide truncated by the pml . in this section
, we will derive pml models that truncate the computational domain in the @xmath56direction using two standard approaches , the complex coordinate stretching technique @xcite and berenger s splitting method @xcite .
the two approaches are mathematically analogous but yield two different formulations of the pml , the unsplit pml and split field pml , respectively .
the difference lies on the choice of variables in the time domain .
as we will see below , by appropriate choice of variables the split - field @xcite can be formulated as the unsplit @xcite .
this shows that the general solutions of the two pml formulations are identical .
let the laplace transform of @xmath110 be defined by @xmath111 in order to derive the pml we assume homogeneous initial data and take the laplace transform of the maxwell s equations in time , having @xmath112 here @xmath113 , with the positive real part @xmath114 , is the dual variable to time .
note that we have tacitly assumed homogeneous initial data
. a general approach to deriving a pml was introduced in @xcite .
the main idea is to analytically continue the maxwell s equation to a complex spatial coordinate system where all spatially oscillating solutions are turned into exponentially decaying solutions . by using the complex coordinate stretching technique
it is straightforward to construct the pml for many wave equations .
however , in order to localize the pml in time auxiliary variables are often introduced , to avoid explicit convolution operations .
a pml truncating the boundary in the @xmath56direction can be derived by considering the , laplace transformed , maxwell s equation and applying the complex coordinate transformation @xmath115 we have [ eq : maxwell_laplace_pml_unsplit ] @xmath116 where @xmath117 is a pml complex metric . in this study , we will use the complex metric @xmath118 corresponding to the standard pml model . here
, @xmath119 is a positive real function that vanishes in @xmath120 and @xmath113 is the dual variable to @xmath121 .
more complicated complex metrics can be derived by using a modal ansatz @xcite , constructed to support only spatially decaying modes .
the next step is to invert the laplace transforms in back to the time domain . to do this
we multiply equations by @xmath117 and introduce the auxiliary variable @xmath122 inverting the laplace transforms yields the time dependent pml model [ eq : maxwell_pml_split_2 ] @xmath123 using instead the auxiliary variable @xmath124 and inverting the laplace transform yields the time dependent pml model [ eq : maxwell_pml_waveguide ] @xmath125 we can also choose auxiliary variables to obtain the physically motivated pml used in @xcite . to begin , we consider and define the auxiliary variables @xmath126 inverting the laplace transform and omitting the tilde signs ( @xmath127 ) results in the physically motivated unsplit pml of @xcite [ eq : maxwell_pml_ababarnel ] @xmath128 note that the pml was originally derived using physical arguments @xcite .
there are certainly other ways to choose auxiliary variables , see for example @xcite .
however , all resulting pml models can be shown to be linearly equivalent to , and . since in the laplace space the pml models , , correspond to , a complex change of coordinates , it follows that the equations are perfectly matched to the maxwell s equation by construction @xcite .
berenger s original idea @xcite was to split the electric field and the corresponding equations into two artificial tangential components before special lower order terms that simulate the absorption of waves are added .
that is , consider the maxwell s equation in the laplace space .
introduce the split variables @xmath129 and equations [ eq : maxwell_laplace_pml_split ] @xmath130 where @xmath119 is a positive real function that vanishes in @xmath120 .
note that equation can be derived by splitting the variables and equations , and then introducing the complex coordinate transformation .
this shows that the two approaches are mathematically analogous .
however , unlike the unsplit pml , the split - field pml introduces the additional degrees of before the pml transformation is introduced . it can be shown that the restriction of the general solution to the pml in @xmath120 coincides with the general solution to the maxwell s equation , see @xcite .
therefore , the pml equation is perfectly matched to the maxwell s equation by construction . by inverting the laplace
transforms in we obtain the time - dependent split - field pml @xcite [ eq : maxwell_pml_split_1 ] @xmath131 we will show that the split - field pml can be reformulated as the unsplit pml models , . following @xcite , it is possible to re - introduce the electric field @xmath132 in the pml equations while eliminating one of the split variables . by summing , together and using the identity @xmath133 we again
obtain the pml equation .
we can also rewrite the equation in the form .
multiplying equation by @xmath134 and introducing @xmath135 we obtain .
note that in @xcite the berenger s pml and the modal unsplit pml , in the form , gave identical numerical results .
the turning point is in efficient numerical implementation . implementing the split
field pml also requires the splitting of the electric field in the interior .
the unsplit pml appears as a lower order modification of the maxwell equations and does not require such unphysical splitting of the field variables . in the laplace space , the pml models , , and
are mathematically equivalent @xcite .
the difference in the time domain lies on the choice of auxiliary variables . by standard definitions
the physically motivated pml satisfies the definition of a strongly hyperbolic system . on the order hand , also by standard definitions the modal pmls , and the split
field pml are weakly hyperbolic .
it is a bit strange though to say that the strongly hyperbolic system is equivalent to the weakly hyperbolic problems ,
, . however , as we will see in the next section , it is possible to rewrite the pmls , , as a second order system and derive an energy estimate for the solutions in a certain sobolev norm .
thus , showing well posedness also for the time
dependent pml models , , . in this study , we will investigate discrete stability of three pmls resulting from the modifications of the discrete equations .
the first candidate is an unsplit modal pml recently derived in @xcite and given by .
note that the results obtained for the pml can be easily extended to the pml model . we will not consider the pml model in this study .
the second is the so called physically motivated unsplit pml of @xcite defined by , and
the third candidate is the classical berenger s @xcite split
field pml . here , the damping function is given by the third degree monomial @xmath136 the damping coefficient is @xmath137 , which allows a relative modeling error of magnitude @xmath138 from the outer boundaries , see @xcite .
note that @xmath139 is a user input parameter .
we will now perform some instructive numerical experiments which are the motivations of the analysis in the coming sections .
we consider a typical test problem as in @xcite with @xmath140 in , and we set @xmath141 , @xmath142 , and the pml width is @xmath143 we discretize the domain with a uniform spatial step @xmath144 .
the pml equations are discretized straightforwardly .
that is , we replace the spatial derivative in with a finite difference sbp operator and append the discrete auxiliary functions to the discrete equations , we have [ eq : maxwell_pml1_waveguide ] @xmath145 note that for the pml model , the auxiliary differential equation is governed by an ordinary differential equation .
the corresponding discrete pml is [ eq : maxwell_pml2_waveguide ] @xmath146 the corresponding semi - discretization for the split field pml
is [ eq : maxwell_pml3_waveguide ] @xmath147 the penalty parameters are @xmath148 similarly , as in the continuous case , using the identity @xmath149 we can eliminate the auxiliary variable @xmath150 and rewrite the split
field pml exactly as the discrete modal pml . therefore , we expect the pml models and to have identical stability properties
. note that when the damping vanishes @xmath151 we can eliminate the auxiliary variables in , and the split variable in to obtain . since the discrete approximation of the interior , when @xmath151 , is stable , we hope that the discrete approximations , and
, with @xmath152 , corresponding to the pml are also stable .
we set the smooth initial data @xmath153 for the electric field and use zero initial data for all other variables .
high order accurate sbp operators of interior order of accuracy 2 , 4 , 6 are used to approximate spatial derivatives .
we advance the solutions in time for the three pml models using the classical fourth order accurate runge
kutta scheme .
the final time is @xmath154 and the time step is @xmath155 .
here , the modeling error is @xmath156 .
the time histories of the discrete @xmath157-norm of the electric field are recorded in figures [ fig : model_standard_ourpml ] , [ fig : model_standard_ababarnel ] and [ fig : model_split_field ] for the discrete pml models , and , repsectively .
clearly , from figures [ fig : model_standard_ourpml ] , [ fig : model_standard_ababarnel ] and [ fig : model_split_field ] , the three discrete pml models , and are unstable for sbp operators of accuracy 4 and 6 .
note that the numerical solution for higher order accurate schemes are more sensitive to numerical instability . for the chosen simulation duration , @xmath154 ( 12 500 time steps )
, we did not observe any growth for the second order accurate case , see also figures [ fig : model_standard_ourpml ] , [ fig : model_standard_ababarnel ] and [ fig : model_split_field ] .
it is also important to see that the stability behavior of discrete modal pml and discrete split
field pml are identical . however ,
when the absorption function is absent ( @xmath151 ) the solutions , see figure [ fig : model_standard ] , decay throughout the simulation for all orders of accuracy .
this is consistent with the theory , since the numerical scheme , , of the interior is provably stable .
note also when @xmath151 , the corresponding ode is not stiff .
[ h ! ] as in @xcite , it is possible to argue here that the growth seen for the 4th and 6th order accurate schemes are due to instabilities in the continuous pml models .
however , in @xcite , stability analysis of an ivp corresponding to was performed in detail .
the result is that the ivp is asymptotically stable .
we believe that the instability seen here and , probably , some of the growth observed in @xcite are caused by inappropriate numerical boundary treatments in the pml . to investigate further ,
we have also performed numerical experiments with smaller time steps or / and spatial steps . in some cases
mesh refinement or smaller time step eliminates the growth , but in other cases the instabilities persist , but it occurs at much later times .
for instance , using half of the time step , @xmath158 , postpones the initiation of growth until about @xmath159 for the pml models and with a 4th order accurate sbp operator . while in all other cases growth was not seen for the entire simulation duration .
these simulations are a strong indication that the growth observed here is a numerical artifact .
in particular , for the discrete physically motivated pml , figure [ fig : model_standard_ababarnel_2dt02h ] indicates that the growth seen in figure [ fig : model_standard_ababarnel ] is due to numerical stiffness .
thus , the numerical approximation of the pml may not be suitable for efficient explicit time integration .
these results are also consistent with some of the conclusions drawn in @xcite , which show that growth for pmls depends on both the discretization used and the pml model . [ h ! ] in the next section , we will show that the ibvps , , , and with , are also stable . to enable a systematic development of stable numerical boundary procedures
we will derive continuous energy estimates . in section 5 , by mimicking the continuous energy estimates we will design accurate and stable numerical boundary treatments for the pmls , , , and , subject to the boundary conditions , .
the analysis in @xcite shows that the constant coefficient cauchy pml problem is asymptotically stable . since the continuous pml models , , and are equivalent , the results in @xcite for the cauchy problem hold also for the pml models , and . note that the boundary condition extends into the pml .
in addition the pml is terminated in the @xmath56direction by the boundary condition .
the stability analysis of , , or
subject to the boundary conditions and is not as straightforward . here , we prove that the pml , , or with the boundary conditions , is stable .
first , we will use a modal analysis for a constant coefficient pml .
second , we will derive energy estimates for the variable coefficients case . to perform a modal analysis
, we assume constant coefficients .
we split the pml problem into three : 1 ) a cauchy pml problem with no boundary conditions , 2 ) a lower half plane pml problem , @xmath160 , with the boundary condition at @xmath161 , and 3 ) a left half - plane pml problem , @xmath162 , with the boundary condition at @xmath163 .
each of the three pml problems above can be analyzed separately .
we note that the lower half plane pml problem and the left half
plane pml problem are distinct and must be analyzed severally .
the cauchy problem can be analyzed , as in @xcite , using standard fourier methods .
the analysis in @xcite shows that the constant coefficient cauchy pml problem , , or
is asymptotically stable . as before we have [ theo : cauchy_problem ] the constant coefficient cauchy pml problem , , or with @xmath164 is asymptotically stable .
consider now the constant coefficient pml
, , or in a lower half plane , @xmath160 , with the boundary condition at @xmath161 .
note that the relevant ansatz is no longer a pure fourier mode , but a simple wave solution on the form @xmath165 in order to prove the stability of the pml , , or
subject to the boundary condition , we will show that the only solution of the form with @xmath166 for any @xmath167 is the trivial solution , @xmath168 . to begin with ,
we introduce the complex number @xmath169 and define the branch cut of @xmath170 by @xmath171 next , we introduce in the pml , , or , and the boundary condition
we can eliminate the magnetic fields and the auxiliary variable , then we obtain a second order ode , @xmath172 subject to the boundary condition @xmath173 or the pec boundary condition @xmath174 consider now @xmath175 , we can construct modal solutions , for , of the form @xmath176 where @xmath177 and @xmath178 are parameters which must not be zero at the same time .
note that it can be shown that @xmath179 for all @xmath180 , see lemma [ lem : duru1 ] in the appendix .
thus , boundedness of the solutions at @xmath181 implies @xmath182 .
the parameter @xmath183 is determined by the boundary condition .
introducing in we obtain @xmath184 since @xmath185 , we have nontrivial solutions only if @xmath186 the following lemma characterizes the roots of the dispersion relation .
[ lem : lower_half_plane ] consider the dispersion relation @xmath187 defined in with @xmath188 the equation @xmath189 has no solution for all @xmath190 , @xmath191 and @xmath192 consider first @xmath151 , we have @xmath193 , with the solution @xmath194 . note that since the maxwell s equation subject to the boundary conditions satisfy the energy estimate we must have @xmath195 , for all @xmath196 .
otherwise if @xmath197 , then the energy will grow for some solutions , which is in contradiction with the energy estimate . by inspection ,
the zero root @xmath198 is also not a solution of @xmath199 if @xmath167 , then the equation @xmath189 has the solution @xmath200 therefore , any @xmath167 will move all roots @xmath113 further into the stable complex plane . by lemma [ lem : lower_half_plane ]
we must have @xmath201 for all @xmath167 , @xmath202 .
this also implies that @xmath203 in for all @xmath204 .
therefore , the trivial solution is the only possible solution of with or with , of the form with @xmath205 .
consider now the pec boundary condition and substitute , with @xmath206 , we have @xmath207 thus the pec boundary condition supports only trivial solutions , @xmath208 .
consider now the constant coefficient pml , , or in the left half
plane , @xmath210 @xmath211 , with the boundary condition at @xmath212 .
we make the ansatz @xmath213 in order to prove the stability of the pml , , , or
, subject to the boundary condition , we will show that there are no nontrivial solutions on the form with @xmath214 for any @xmath152 .
we insert in , , or and in the boundary condition . as
before , we eliminate the magnetic fields and the auxiliary variable to obtain a second order ordinary differential equation , @xmath215 subject to the boundary conditions @xmath216 or the pec boundary condition @xmath217 consider now @xmath175 , we can construct modal solutions , for , on the form @xmath218 where @xmath219 and @xmath178 are parameters which must not be zero at the same time .
note that it can be shown that @xmath179 for all @xmath180 , see lemma [ lem : duru2 ] in the appendix .
thus , boundedness of the solution at @xmath220 implies @xmath182 .
the parameter @xmath183 is determined by the boundary condition .
introducing in we obtain @xmath221 since @xmath185 , we have nontrivial solutions only if @xmath222 it is particularly important to note that the dispersion relation @xmath223 , defined by , is independent of the pml parameter @xmath164 .
the dispersion relation corresponds to that of the undamped problem , and it is not perturbed by the pml .
this is unlike the dispersion relation , @xmath224 for the lower half plane pml problem , which depends on the pml parameter @xmath164 . since
, the maxwell s equation subject to the boundary conditions and satisfy the energy estimate , the following lemma characterizing the roots of the dispersion relation holds [ lem : left_half_plane ] consider the dispersion relation @xmath223 defined in with @xmath225 .
the equation @xmath226 has no solution @xmath113 for all @xmath190 and @xmath227 . since the maxwell s equation with the boundary condition at @xmath163 satisfy the energy estimate we must have @xmath228 , for all @xmath227
otherwise if @xmath229 , then the energy will grow for some solutions , which is also in contradiction with the energy estimate . again by inspection , the zero root @xmath230 is not a solution of @xmath231 note that if @xmath232 we have @xmath233 .
next , we consider the case @xmath234 , @xmath235 , and introduce @xmath236 @xmath237 if @xmath238 , then we have @xmath239 since the square root can never be negative . if @xmath240 , then we also have @xmath239 since the square root is zero or purely imaginary .
therefore , we must have @xmath241 , for all @xmath227 .
this completes the proof of the lemma . as before , for @xmath205
the trivial solution is the only possible solution of with or with , on the form .
consider now the pec boundary condition and substitute , with @xmath206 , we have @xmath207 the pec boundary condition also supports only trivial solutions , @xmath208 . by theorem [ theo :
cauchy_problem ] , and lemmas [ lem : lower_half_plane ] and [ lem : left_half_plane ] , it then follows that all non trivial modes in the pml decay .
the difficulty lies in constructing accurate and stable numerical approximations for the pml , , , or
subject to the boundary conditions and , and ensuring numerical stability .
consider now the constant coefficient pml , , or in the rectangular domain , @xmath242 , and the boundary conditions , .
we will prove that the corresponding constant coefficient ibvp can not support growing modes .
to begin , we assume homogeneous initial data and take the laplace transform of the pml equations and boundary conditions in time . we can eliminate the magnetic fields and all the auxiliary variables , we have @xmath243 subject to the boundary conditions @xmath244 @xmath245 or @xmath246 @xmath247 we define the quantity @xmath248 for the pec boundary conditions , , the boundary terms vanish and we have @xmath249 note that @xmath250 , see lemma [ lem : duru3 ] in the appendix .
therefore , @xmath251 and @xmath252 are energies .
note also that for @xmath253 , @xmath254 we have @xmath255 and @xmath256 we can now prove the theorem : [ theorem : wellposedness_laplace ] for any @xmath113 , with @xmath257 , and @xmath258 all solutions @xmath259 of the constant coefficient pml in the rectangular domain , with the boundary conditions , or , must satisfy @xmath260 multiply equation from the left by @xmath261 , and integrate over the whole domain , we have @xmath262 integrating by parts and using the boundary conditions , obtaining @xmath263 we add the complex conjugates of the products in to obtain @xmath264 note that in particular if the pec boundary conditions , are imposed , the boundary terms vanish and @xmath265 thus , for @xmath114 no nontrivial solutions @xmath259 can be supported by the pml . to enable the development of a systematic and rigorous stable numerical boundary procedure for the time
dependent pml we will derive energy estimates for the variable coefficient pml .
the importance of the energy estimates are twofold : the first is that the energy estimates establish the well posedness of the variable coefficients pmls .
the second is that by mimicking the energy estimates we can construct high order accurate , stable and convergent discrete approximations of the pml in a bounded domain .
the first step is to rewrite the pml and the boundary conditions , as a second order system . differentiating equation with respect to time and using equations to eliminate the time
derivatives on the right hand side , we obtain [ eq : maxwell_pml_waveguide_2nd2 ] @xmath266 the corresponding boundary conditions are @xmath267 @xmath268 or @xmath269 @xmath270 here , the length and width of the computational domain are @xmath8 and @xmath9 , respectively . to derive an energy estimate we introduce the relevant energy norms @xmath271 and @xmath272 note that away from the boundaries , the energy norm is analogous to the norms derived in @xcite .
the difference between and @xcite is that the energy norm introduced here contains also boundary norms , @xmath273 , @xmath274 .
note also that @xmath275 , @xmath276 , if and only if @xmath277 when the damping vanishes , @xmath151 , we have the physical energy norms @xmath278 and @xmath279 let us define the spaces of real functions @xmath280 @xmath281 and @xmath282 the space of square integrable real functions .
it can be shown that @xmath283 and @xmath284 we can now prove the theorem , see appendix [ proof : wellposedness ] : [ theorem : wellposedness ] all solutions of the perfectly matched layer or ( [ eq : maxwell_pml_waveguide_2nd])([eq : maxwell_pml_waveguide_y ] ) subject to the boundary conditions ( [ eq : wall_bc_y1 ] ) at @xmath11 and ( [ eq : wall_bc_x1 ] ) at @xmath285 , or ( [ eq : wall_bc_y1_pec ] ) at @xmath11 and ( [ eq : wall_bc_x1_pec ] ) at @xmath285 satisfies the energy estimate @xmath286 where @xmath287 it is particularly noteworthy that the pml models and can be rewritten directly in the time domain as . therefore , theorem [ theorem : wellposedness ] holds also for the solutions of and . in the next section , by mimicking the continuous energy estimate , we will develop stable numerical boundary procedures for the pml models , and , subject to the boundary conditions ( [ eq : wall_bc_y1 ] ) at @xmath11 and ( [ eq : wall_bc_x1 ] ) at @xmath288 . the physically motivated pml model , satisfies the definition of a strongly hyperbolic system . by applying the energy method directly to the pml , , subject to the boundary conditions , , with @xmath36
, we have @xmath289 in particular if @xmath38 in , or @xmath18 ( we consider the pec condition ) in , then @xmath290 note that all terms , excepting @xmath291 , in the right hand sides of , are dissipative .
we denote the energy by @xmath292 and @xmath293 if @xmath38 in , or @xmath18 ( we consider the pec condition ) in .
we have [ theorem : wellposedness_physical ] the solutions of the perfectly matched layer
subject to the boundary conditions , with @xmath294 or @xmath295 satisfy the energy estimate @xmath296 where @xmath287
the main focus of this section is to design accurate , stable and efficient numerical boundary procedures for the pmls , , , and , subject to the boundary conditions , .
we will use sbp finite difference operators , see @xcite , to approximate the spatial derivatives and impose boundary conditions weakly using penalties @xcite . to ensure numerical stability
we will impose boundary conditions in a manner which allows a derivation of discrete energy estimates analogous to the continuous energy estimate ( [ eq : energyestimate_continuous ] ) .
for the physically motivated pml model , an efficient numerical boundary treatment is achieved by using characteristics and reducing the strength of the penalty parameters .
finally , we will show that the corresponding discrete models do not support growing modes . in standard numerical methods @xcite , once the interior scheme is proven stable the discrete pml
is obtained by discretizing the auxiliary differential equations and appending the auxiliary functions accordingly .
usually no more attention is placed on how the boundary conditions are implemented in the pml .
note that numerical experiments presented in section 3 and the next section show that a straightforward enforcement of the boundary conditions , and , can lead to unstable solutions in the pml . a stable discrete approximation for the modal pml
can be obtained by extending the enforcement of the boundary condition to the auxiliary differential equation , having [ eq : maxwell_pml1_waveguide_stable_0 ] @xmath297 note that we have included the term involving @xmath298 . the standard choice @xmath299 , which corresponds to
, is obtained by simply replacing spatial derivatives with difference operators in the pml .
the other choice , @xmath300 , extends weak enforcements of the boundary conditions , in the @xmath57-direction , to the auxiliary differential equation .
it is particularly important to note that this extension , with @xmath300 , does not destroy the accuracy of the scheme . when @xmath151 , the auxiliary variable @xmath301 also vanishes , and the discrete approximation , for any value of @xmath298 ,
correspond to the stable discrete approximation . however , as we will see later , when @xmath167 the particular choice @xmath302 ensures numerical stability .
to be able to simplify the analysis and also make comparisons with results published in the literature , we will focus on the setup considered in section [ sect : num_example ] and @xcite .
therefore , we set @xmath140 and the penalty parameters @xmath303 having [ eq : maxwell_pml1_waveguide_stable ] @xmath304 we will derive a discrete energy estimate analogous to the continuous energy estimate ( [ eq : energyestimate_continuous ] ) .
first and foremost , we rewrite as a second order system .
differentiate equation in time , and use equations to eliminate the time derivatives in the right hand side .
we have [ eq : maxwell_pml1_waveguide_stable_step1 ] @xmath305 subsequently , by using the sbp property we eliminate some of the sat terms and obtain [ eq : maxwell_pml_waveguide_discrete_first ] @xmath306 introduce the discrete energy @xmath307 when @xmath151 we have the standard discrete energy @xmath308 we can also prove the following , see appendix [ proof : discrete_stability ] . [
theorem : discrete_stability ] consider the discrete approximation or of the the perfectly matched layer
subject to the boundary conditions ( [ eq : wall_bc_y1 ] ) at @xmath11 and ( [ eq : wall_bc_x1 ] ) at @xmath285 .
if @xmath302 , then the quantity @xmath309 defined in is an energy and the solutions of the discrete system or satisfy the energy estimate @xmath310 note that when @xmath151 we have a strict energy estimate @xmath311 for any values of @xmath298 . however , for @xmath152 , if @xmath312 we are unable to derive a discrete energy estimate for the discrete pml . in standard numerical codes , as shown in section 3 , once the interior scheme is proven stable , the pml is then added as lower order modification , leading to the standard choice @xmath299 .
if @xmath313 , then there is a non - vanishing boundary term , @xmath314 the boundary term will probably diminish with mesh refinement , @xmath315 .
however , on a realistic mesh the boundary term may be nontrivial .
the energy estimate above does not exclude the possibility of exponentially growing solutions for @xmath167 .
however , the estimate can be extended to the error equations to proof convergence of the solutions of the discrete approximation or for any finite time interval @xmath316 $ ] .
below , we will prove that at constant coefficients the semi
discrete pml problem or can not support exponentially growing solutions .
consider now the semi discrete system and take the laplace transform in time , we have [ eq : maxwell_pml_waveguide_discrete_first_laplace ] @xmath317 here , @xmath113 with @xmath114 , is the dual variable to time . note that in we have also used the sbp properties and to eliminate some of the sat terms
. we can also eliminate the magnetic fields in using , and obtain @xmath318 where @xmath319 .
we define the discrete quantity @xmath320 note that it can be shown that @xmath321 see lemma [ lem : duru3 ] in the appendix . therefore , if @xmath322 then @xmath323 is an energy .
we can now prove the following result establishing the stability of the discretization .
[ theo : asymptotic_stability ] consider the constant coefficient semi
discrete pml with @xmath114 , @xmath152 and @xmath302 .
all solutions of satisfy the energy equation @xmath324 multiply equation from the left by @xmath325 and add the complex conjugate of the product we have @xmath326 on the left hand side of we recognize the energy @xmath327 . on the right hand side , if @xmath302 , then all the expressions vanish , yielding @xmath328 the proof of the theorem is complete . by theorem
no nontrivial solution of the discrete problem can grow exponentially for all @xmath152 and @xmath329 however , for @xmath152 and @xmath313 there are indefinite non vanishing boundary terms . as we saw in the previous experiments these terms can ruin the accuracy of numerical solutions . here ,
we develop a stable numerical method for the split
field pml . since the split
field pml can be rewritten as the modal unsplit , our task is made simpler .
we will design a numerical method for such that the corresponding discrete problem can be rewritten as the stable discrete modal pml with @xmath302
. we can then apply theorem [ theorem : discrete_stability ] and theorem [ theo : asymptotic_stability ] to prove the stability of the discretization .
the corresponding stable semi - discretization for the split
field pml is [ eq : maxwell_stablesplitfieldpml ] @xmath330 note that we have moved the sat - term @xmath67 to the last equation .
this is important since we can rewrite the discrete approximation as the stable discrete unsplit pml with @xmath302 .
similarly , as in the continuous case , using the identity @xmath149 we can eliminate the split variable @xmath150 , obtaining [ eq : maxwell_stablesplitfieldpml_modal ] @xmath331 note that equation above corresponds to discretizing the unsplit pml . now multiply equation by @xmath332 and introduce @xmath333
, we obtain exactly the stable discrete unsplit pml with @xmath302
. we can now apply theorem [ theorem : discrete_stability ] and theorem [ theo : asymptotic_stability ] to prove the temporal stability of the discretizations and .
these results are clearly demonstrated by the numerical experiments performed in the next section .
now , we turn our attention to the physically motivated pml model . we know that the principal part of this pml is symmetric hyperbolic .
the lower order terms appear in the pml in a nontrivial way . as was shown in @xcite ( and we have already seen in section [ sect : num_example ] )
, the pml terms can pose a numerical challenge .
however , we will show that using the sbp sat scheme proposed in section [ sec : sbp_sat ] , we can choose penalties such that any stable numerical boundary procedure for the undamped problem ( with @xmath334 ) satisfying an energy estimate also leads to a discrete energy estimate , analogous to continuous estimate , for the discrete pml . however , the continuous energy for the pml and the corresponding discrete energy for the discrete pml can grow exponentially in time . by theorem , we also know the constant coefficient pml model subject to the boundary conditions ( [ eq : wall_bc_y_0 ] ) , ( [ eq : wall_bc_x ] ) can not support exponentially growing solutions . therefore , at constant coefficients , for a particular set of penalty parameters , we will show that the solutions for the discrete pml can not grow exponentially in time . to begin , consider the discrete pml .
by applying the energy method directly to , we have @xmath335 the boundary terms @xmath76 are exactly the same as the boundary terms derived for the undamped problem in section [ sec : sbp_sat ] .
as before , if @xmath77 and @xmath78 , then we can choose @xmath79 and @xmath336 we have @xmath337 . also , for any @xmath93 and @xmath94 , then we can choose @xmath95 obtaining @xmath337 .
note that all terms in the right hand side of , excepting @xmath338 , are dissipative . introducing the discrete energy @xmath339 we have [ theorem : discrete_stability2 ]
consider the discrete approximation for the physically motivated pml
subject to the boundary conditions ( [ eq : wall_bc_y_0 ] ) at @xmath11 and ( [ eq : wall_bc_x ] ) at @xmath340 . for * @xmath77 and @xmath78 , and @xmath79 with @xmath336 or * @xmath93 and @xmath94 , and @xmath341 then @xmath337 and the quantity @xmath342 defined in is an energy . for these sets of penalty parameters
the solutions of the discrete pml problem satisfy the energy estimate @xmath343 by , the discrete energy can grow exponentially in time .
however , the modal analysis presented in section [ section : modal_analysis ] shows that the continuous pml problem subject to the boundary conditions ( [ eq : wall_bc_y ] ) at @xmath11 and ( [ eq : wall_bc_x ] ) at @xmath340 does not support temporally growing modes .
therefore , in the discrete setting , it will be more desirable to design numerical methods that ensure that all eigenvalues of the discrete spatial operator satisfy @xmath344 for any @xmath345 .
this has the potential to eliminate any nonphysical temporal growth . at constant coefficients
we will show that the corresponding discretization used in section [ sect : num_example ] can not allow growing modes consider the discrete approximation with @xmath140 and @xmath346 we have [ eq : maxwell_pml2_waveguide_stable ] @xmath347 after laplace transformation of in time , we can eliminate the auxiliary variable and the magnetic fields obtaining @xmath348 where @xmath319 and @xmath114 . as in , note
also that we have used the sbp property to eliminate some of the sat boundary terms .
we define the corresponding discrete energy @xmath349 note that by lemma [ lem : duru3 ] in the appendix , we have @xmath350 therefore , @xmath351 is an energy
. we can now prove the following result establishing the stability of the discretization .
[ theo : asymptotic_stability_abarbanel ] consider the constant coefficient semi
discrete pml with @xmath114 , @xmath152 .
all solutions of satisfy the energy equation @xmath352 the proof of theorem [ theo : asymptotic_stability_abarbanel ] is completely analogous to the proof of theorem [ theo : asymptotic_stability ] .
we therefore omit it here .
note that theorem [ theo : asymptotic_stability_abarbanel ] shows that the discrete pml problem can not have eigenvalues with positive real parts .
however , for efficient explicit time integration we do not only need the temporal eigenvalues to have non positive real parts , the spectral radius , @xmath353 , must also be sufficiently small .
note that in section [ sect : num_example ] , the penalty parameters @xmath354 when combined with the physically motivated pml yielded unstable solutions for the time step @xmath355 .
however , when we reduced the time step to @xmath158 the numerical solutions are stable .
this shows that the above penalty parameters for the physically motivated pml is not adequate for efficient explicit time integration .
it is possible though to construct stable boundary procedures which will not restrict the time step further when the pml is introduced . for the physically motivated pml we propose to use the stable penalty parameters , @xmath356 note that the maximum weights of these penalty parameters are smaller . for many choices of boundary parameters @xmath357 , this has the potential to reduce numerical stiffness in the discrete pml .
numerical experiments performed in the next section confirm this .
in this section , we present some numerical tests to verify the analysis of previous sections . in particular , we will investigate temporal stability of the discrete pmls , ,
, , and also evaluate the accuracy of the pml , in an electromagnetic waveguide . to begin with , we will repeat the experiments of section 3 .
we consider exactly the same setup as in section 3 .
also , the same discretization parameters , time step @xmath155 , absorption function and damping coefficient are used .
we run the simulations now for a very long time , until @xmath358 ( 125 000 time steps ) , using the discrete pmls , with @xmath302 , with @xmath359 and , respectively .
the time histories of the @xmath157-norm of the solutions are recorded in figures [ fig : model1_stable][fig : model3_stable ] for the discrete pmls , and , respectively .
clearly , the solutions decay with time throughout the simulation .
this verifies the analysis of sections 4 and 5 .
the numerical growth seen in section 3 are due to inappropriate boundary treatments in the pml .
we also believe that some of the growth seen in @xcite are probably caused by numerical enforcements of boundary conditions .
it is noteworthy that the behaviors of the discrete pmls , and , are completely analogous . [ h ! ] we have also performed numerous numerical experiments with different boundary parameters , @xmath360 and penalty parameters @xmath361 .
unfortunately , because of space , we can not report all of them here .
it is particularly important to note that applying the penalty @xmath359 to the discrete modal pml
, with @xmath299 , yields unstable solutions .
numerical growth was also observed when the stable boundary treatment used in was applied to the physically motivated pml without decreasing the time step further .
this indicates that each pml model must be treated individually . a stable and efficient numerical boundary treatment for a particular pml model may not necessarily yield stable and efficient numerical scheme for another pml modeling the same physical phenomena .
of course , when the underlying physical model changes , the corresponding pml model will also change . we will also expect different stable numerical boundary treatments . in the next section
, we will verify the stability and accuracy of the pml in an electromagnetic waveguide . here
, we will focus on the discrete pml . the results obtained for the discrete pmls , and , are analogous , and will not be presented .
consider the rectangular semi
infinite electromagnetic waveguide @xmath362 , @xmath363 . at the upper wall , @xmath364 of the waveguide ,
the boundary condition is a magnetic source @xmath365 , where @xmath366 and @xmath367 at the lower wall , @xmath368 the boundary condition is homogeneous @xmath369 . at the left wall , @xmath370 , we impose the characteristic boundary condition @xmath371 . in order to perform numerical simulations ,
we truncate the right boundary at @xmath372 with a pml of width @xmath373 .
we terminate the pml at @xmath374 with the characteristic boundary condition @xmath375 .
homogeneous initial conditions @xmath376 are used for all variables .
we also use the cubic monomial damping profile @xmath377 with the damping coefficient @xmath378 , where the factor @xmath379 is empirically determined and @xmath380 is the spatial step .
we approximate all spatial derivatives using sbp operators with interior order of accuracy 2 , 4 , 6 .
boundary conditions are imposed weakly using penalties .
as before , we discretize in time using the classical 4th order accurate runge
kutta method . in the first sets of experiments we investigate discrete stability .
consider sbp finite difference operators of order of accuracy 2 , 4 , 6 , respectively .
we use the spatial step @xmath381 and the time step @xmath355 , and evolve the solutions until @xmath382 ( 250 000 time steps ) with @xmath383 , respectively , in .
note that here the domain , spatial and time steps are much smaller than the ones used in the previous sections .
the snapshots of the solutions at @xmath384 are shown in figure [ fig : slution_t300 ] , for the 4th order accurate sbp operator .
note that initially the two solutions , using @xmath383 , are similar . for the standard choice
, @xmath299 , the solution in the pml starts growing after a longtime , just before @xmath385 , and spreads into the computational domain as time passes . at @xmath384 , growth in the pml
has already corrupted the solutions everywhere .
the 2nd and 6th order accurate operators with @xmath299 are also unstable , but with slower growth rates . on the other hand , for @xmath386 the solutions decay throughout the simulation for all sbp operators .
it is important to remark that the observed growth is due to the unstable boundary treatments in the pml .
the conclusion here is that a numerical method which is stable in the absence of the pml , when @xmath151 , can result in an unstable scheme when , @xmath167 , the pml is included .
it is possible though to design stable numerical methods which can be proven stable for all @xmath387 here , numerical stability for the discrete pml was achieved by using @xmath302 in equation without any additional cost or difficulty .
next , we evaluate the accuracy of the pml .
we set @xmath302 and compute the solutions for various resolutions using sbp operators of interior order of accuracy 2 , 4 and 6 , until the final time , @xmath388 the snapshots of the solutions at @xmath389 are displayed in figure [ fig : maxwell_solutions_waveguide ] , showing the evolution of the electric field on the boundary , the propagation and reflection of waves in the waveguide , and the absorption of waves by the pml .
the errors introduced when a pml is used in computations can be divided into two different categories : numerical reflections and the modeling error .
numerical reflections are discrete effects introduced by discretizing the pml and seen inside the computational domain . the modeling error is introduced because the layer and the magnitude of damping coefficient are finite .
numerical reflections should converge to zero as the mesh is refined .
note that the modeling error is not caused by numerical approximations , and thus is independent of the numerical method used .
the modeling error is expected to decrease as the pml width or the magnitude of damping coefficient increases . for sufficiently small mesh
sizes numerical reflections are infinitesimally small and the modeling error dominates the total pml error . in order to measure numerical errors , we also computed a reference solution in a larger domain without the pml . by comparing the reference solutions to the pml solutions in the interior , measured in the maximum norm ,
we obtain accurate measure of total pml errors .
the numerical errors at @xmath390 are displayed in table [ tab : error_pml_waveguide ] , and the time history of the error are shown in figure [ fig : pmlerror_waveguide ] .
note that the errors converge to zero exponentially . because we have used exactly the same damping strength , @xmath391 , for the 2nd
, 4th and 6th order accurate schemes we expect to have the same modeling error independent of the accuracy of the scheme used .
it is also important to note that for small mesh sizes the modeling error dominates and
the behavior of the total pml error does not depend on the order of accuracy of the scheme .
.total pml errors and the convergence of the error in a waveguide using sbp operators of interior order 2 , 4 , 6 for different grid resolutions . [ cols="^,^,^,^,^,^,^,^ " , ] [ h ! ]
in this paper , we have studied the ( long time ) temporal energy growth associated with numerical approximations of the pml in bounded domains .
three different unsplit pml models and the classical split
field pml model are considered .
we proved the stability of the constant coefficient ibvp for the pml .
we also derived energy estimates for the constant coefficient ibvp in the laplace space and for variable coefficients pml in the time domain domain .
thus , establishing well posedness and stability of the ibvp corresponding to the pml .
we have demonstrated in theory and numerical experiments that a numerical method which is stable in the absence of the pml can become unstable when the pml is introduced . here , the discrete stability of the pml depends only on the numerical implementation of boundary conditions ; the interior approximations do not play any roles .
we developed high order accurate and stable numerical approximation using sbp finite difference operators to approximate spatial derivatives and weak enforcement of boundary conditions using penalties . by constructing analogous discrete energy estimates
we obtain a bound on the energy growth of the solution at any future time .
further , we showed that the corresponding constant coefficient pml problem can not support temporally growing solution .
we presented numerical experiments demonstrating high order accuracy and longtime stability of the pml .
we note that a stable numerical boundary treatment for a particular pml model may not necessarily yield stable numerical solutions for another pml modeling the same physical phenomena . when the underlying physical model changes , the corresponding pml model will be different .
we also expect different stable numerical boundary treatments .
although the analysis here focuses on the maxwell s equations the results obtained can be straightforwardly applied to equations of acoustics .
we are currently extending the techniques described in this paper to other numerical methods such as finite element methods , spectral element methods , dg finite element / volume methods and mixed spectral - finite element methods .
however , these require substantially more details and analysis .
our preliminary results are quite promising and will be reported in another paper @xcite .
it will be interesting to compare the results obtained from a dg method to the results obtained in this paper using high order finite difference sbp operators .
introduce the complex number @xmath392 and define the branch cut of @xmath170 by @xmath393 the following lemma was adapted from lemma 6 in @xcite . [
lem : heinz_anders ] let @xmath394 be a real number and let @xmath395 be a complex number where @xmath396 . consider the relation @xmath397 there are positive real numbers @xmath398 , @xmath399 such that @xmath400 , @xmath401 .
we can also prove the following lemma [ lem : duru1 ] let @xmath152 , @xmath394 be real numbers and let @xmath395 be a complex number where @xmath396 .
consider the relation @xmath402 there are positive real numbers @xmath403 , @xmath404 such that @xmath405 consider @xmath406 [ lem : duru2 ] let @xmath152 , @xmath394 be real numbers and let @xmath395 be a complex number where @xmath396 .
consider the relation @xmath397 there are positive real numbers @xmath403 , @xmath404 such that @xmath407 consider @xmath408 [ lem : duru3 ] let @xmath152 , be a nonnegative real number and @xmath395 a complex number where @xmath396 .
consider the relations @xmath409 we have @xmath410 consider @xmath411 note that @xmath412 thus , we have @xmath413 consider now @xmath414 the proof of the lemma is complete .
multiply equation ( [ eq : maxwell_pml_waveguide_2nd ] ) by @xmath415 and integrate over the whole domain .
integration by parts yields @xmath416 imposing the boundary conditions ( [ eq : wall_bc_y1 ] ) at @xmath11 and ( [ eq : wall_bc_x1 ] ) at @xmath340 , we arrive at @xmath417 by adding @xmath418 to both sides of equation ( [ eq : energyestimate_first_continuous ] ) we have @xmath419 by using and in the right hand side of ( [ eq : energyestimate_second_continuous ] ) , we obtain @xmath420 where the energy @xmath421 is define in .
note that if the boundary conditions are ( [ eq : wall_bc_y1_pec ] ) at @xmath11 and ( [ eq : wall_bc_x1_pec ] ) at @xmath288 , from we have @xmath422 thus , cauchy schwartz inequality and the facts @xmath423 , @xmath424 , @xmath425 , @xmath426 , @xmath427 and @xmath428 , @xmath429 , @xmath430 , @xmath431 , @xmath427 lead to @xmath432 the proof of the theorem is complete .
multiplying equation with @xmath435 from the left , we have @xmath436 by adding @xmath437 to both sides of equation ( [ eq : discrete_estimate_1 ] ) we have @xmath438 where the energy @xmath439 is defined in . eliminating the time derivatives on the right hand side of using , we obtain @xmath440 as before , cauchy schwartz inequality and the facts @xmath441 @xmath442 lead to @xmath443 @xmath444 99 s. abarbanel , d. gottlieb , on the construction and analysis of absorbing layers in cem , appl .
, 27 ( 1998 ) , 331340 . s. abarbanel , h. qasimov , s. tsynkov , long - time performance of unsplit pmls with explicit second order schemes , j. sci .
, 41 ( 2009 ) , 112 . s. abarbanel , d. gottlieb , j. s. hesthaven , long time behaviour of the perfectly matched layer equations in computational electromagnetics , j. sci .
17 ( 2002 ) , 14 . s. abarbanel and d. gottlieb , a mathematical analysis of the pml method , j. comput . phys . , 134 ( 1997 ) , 357363 .
s. abarbanel and a. chertock , strict stability of high - order compact implicit finite - difference schemes : the role of boundary conditions for for hyperbolic pdes , i , j. comput .
158 , 125 ( 2000 ) .
s. abarbanel , a. chertock and a. yefet , strict stability of high - order compact implicit finite - difference schemes : the role of boundary conditions for for hyperbolic pdes , ii , j. comput .
160 , 6787 ( 2000 ) .
e. abenius , f , edelvik and c. johansson , waveguide truncation using upml in the finite - element time - domain method , tech .
report , department of information technology , uppsala university , issn 14043203 ; 200506 , ( 2005 ) . t. g. amler , k. duru and i. hoteit , stable dg finite element / volume schemes for the perfectly matched layer for maxwell s equation , in preparation , ( 2014 ) j .- p .
brenger , numerical reflection from fdtd - pmls : a comparison of the split pml with the unsplit and cfs pmls , ieee trans on antennas and propag . , 50 ( 2003 ) , 258265 . j .-
brenger , a perfectly matched layer for the absorption of electromagnetic waves , j. comput .
( 114 ) 1994 , 185200 .
m. h. carpenter , d. gottlieb , s. abarbanel , time - stable boundary conditions for finite - difference schemes solving hyperbolic systems : methodology and application to high - order compact schemes , j. comput .
, 111 ( 1994 ) , 220236 .
w. chew and w. weedon , a 3-d perfectly matched medium from modified maxwell s equations with stretched coordinates , micro .
lett . , 7 ( 1994 ) , 599604
. k. duru , a perfectly matched layer for the time
dependent wave equation in heterogeneous and layered media , j. comput .
, * 257 * : 757 - 781 , ( 2014 ) .
k. duru , g. kreiss , boundary waves and the stability of perfectly matched layer i , in proceedings of num . and
math . aspects of waves ( 2013 ) .
s. d. gedney , an anisotropic perfectly matched layer absorbing media for the truncation of fdtd latices , ieee trans . on antennas and propag . , 44 ( 1996 ) , 16301639 . b. gustafsson , high order difference methods for time dependent pde , springer
verlag , berlin heidelberg , 2008 .
b. gustafsson , on the implementation of boundary conditions for method of lines , bit numer math 38 ( 1998 ) , 293314 .
k. o. friedrichs , symmetric hyperbolic linear differential equations , comm .
pure appl .
, 7 ( 1954 ) , 345392 .
t. hagstrom , new results on absorbing layers and radiation boundary conditions , topics in comput .
wave propag .
notes comput .
( 2003 ) , 142 .
l. halpern , s. petit - bergez , j. rauch , the analysis of matched layers , conflu .
3(2 ) , 159236 ( 2011 ) .
j. hesthaven and t. warburton nodal discontinuous galerkin methods : algorithms , analysis , and applications , springer ( 2008 ) .
jin , w. c. chew , combining pml and abc for the finite - element analysis of scattering problems , microwave and optical technology letters 12 ( 1996 ) , 192197 .
k. kormann , m. kronbichler and b. mller , derivation of strictly stable high order difference approximations for variable - coefficient pde , j. sci comput , doi : 10.1007/s10915 - 011 - 9479 - 1 ( 2012 ) g. kreiss , k. duru , discrete stability of perfectly matched layers for anisotropic wave equations in first and second order formulation , bit numer math ( 2013 ) , doi : 10.1007/s10543 - 013 - 0426 - 4 .
kreiss , and n.a .
petersson , _ boundary estimates for the elastic wave equations in almost incompressible materials , _
siam j. num .
1556 - 1580 , doi : 10.1137/110832847 ( 2012 ) .
kreiss , j. lorenz , initial - boundary value problems and the navier - stokes equations , first ed . , academic press , boston , 1989 .
kreiss , initial boundary value problems for hyperbolic systems , comm .
pure appl .
, 23 ( 1970 ) , 277298 . m. kuzuoglu , r. mittra , frequency dependence of the constitutive parameters of causal perfectly matched anisotropic absorbers , ieee microwave and guided wave letters , 6 ( 1996 ) , 447449 .
lions , j. mtral , o. vacus , _ well - posed absorbing layer for hyperbolic problems , _ numer . math .
( 2002 ) 92 : 535562 .
y. y. lu , minimizing the discrete reflectivity of perfectly matched layers , ieee photonic tech .
18 ( 2006 ) , 487489 . k. mattsson , boundary procedures for summation by parts operators , j. sc . comput . , 18 ( 2003 ) , 133153 . m. motamed , h o .
kreiss , hyperbolic initial boundary value problems which are not boundary stable , numerical analysis , school of computer science and communication , kth , stockholm , sweden ( 2008 ) .
j. nordstrm , error bounded schemes for time - dependent hyperbolic problems , siam j. sci .
comput . , 30 ( 2008 ) , 4659 . h. qasimov , s. tsynkov , lacunae based stabilization of pmls , j. comput
. phys . , 227 ( 2008 ) , 73227345 .
b. sj " ogreen , n. a. petersson , perfectly matched layer for maxwell s equation in second order formulation , j. comput .
phys . , 209 ( 2005 ) , 1946 . b. strand , summation by parts for finite difference approximations for d / dx , j. comput . ,
( 1994 ) 4767 .
m. svrd and j. nordstrm , review of summation - by - parts schemes for initial - boundary - value problems , j. comput .
, 268 ( 2014 ) 1738 .
a. taflove , advances in computational electrodynamics , the finite - difference time - domain artec house inc .
e. turkel and a. yefet , absorbing pml boundary layers for wave like equations , appl .
maths 27 ( 1998 ) , 533557 .
|
in this paper we address the temporal energy growth associated with numerical approximations of the perfectly matched layer ( pml ) for maxwell s equations in first order form . in the literature
, several studies have shown that a numerical method which is stable in the absence of the pml can become unstable when the pml is introduced .
we demonstrate in this paper that this instability can be directly related to numerical treatment of boundary conditions in the pml .
first , at the continuous level , we establish the stability of the constant coefficient initial boundary value problem for the pml . to enable the construction of stable numerical boundary procedures
, we derive energy estimates for the variable coefficient pml .
second , we develop a high order accurate and stable numerical approximation for the pml using summation by parts finite difference operators to approximate spatial derivatives and weak enforcement of boundary conditions using penalties . by constructing analogous discrete energy estimates
we show discrete stability and convergence of the numerical method .
numerical experiments verify the theoretical results .
maxwell s equations , guided waves , boundary waves , normal mode analysis , perfectly matched layers , energy method , well
posedness , stability , high order accuracy , efficiency , finite difference , summation by parts , penalty terms .
35l15 35l05 35l10 35l20 35q61 35q74
|
the experimental results of the past decade indicate that the chiral lagrangian describing ewsb works to one loop accuracy and irrelevant operators ( such as @xmath0 , where @xmath1 are the goldstone bosons eaten by @xmath2 and @xmath3 ) are suppressed by a scale @xmath4 .
this strongly indicates perturbative physics at @xmath5 and a physical higgs boson @xmath6 .
however , this linear sigma model is unstable to quantum corrections leading to the hierarchy problem , meaning that the standard model is in an incomplete description of physics parametrically above the weak scale . to date
, only the mssm provides a solution to the hierarchy problem and weakly coupled physics at 1 tev . in this note
we review models where the higgs is a pseudo - goldstone boson , @xmath7 @xcite , for early attempts see @xcite .
approximate global symmetries , @xmath8 keep the higgs vev small relative to the cut - off , @xmath9 .
the global symmetries are broken by couplings , however , the structure of the symmetry breaking prevents large corrections to the higgs vev . at 100 gev
the model is a two higgs doublet model with a higgs potential identical to the mssm .
there is a mini - desert up to 1 - 3 tev where new particles emerge that cancel the quadratic divergences of the standard model .
the novelty of this set of models is that particles of the same spin cancel the respective quadratic divergences . above the tev scale
there is another mini - desert until the 10 - 30 tev scale where the description of the model becomes strongly coupled and new physics emerges .
however , the standard model is insensitive to this physics .
little higgs models are gauged non - linear sigma model with lagrangians that can be divided into four pieces : @xmath10 the first term is the standard kinetic term for gauge bosons . the second term is the nl@xmath11 m kinetic terms that give mass to some of the gauge bosons .
there is the possibility of adding a scalar potential , often called the plaquette potential , that gives mass to some of the nl@xmath11 m fields .
finally there are the yukawa couplings between the nl@xmath11 m fields and fermions . to study the low energy dynamics of the theory we want to integrate out all the states that are classically massive and have an effective lagrangian in terms of the classically massless fields .
some nl@xmath11 m can be represented by theory spaces where points denote gauge groups , lines are nl@xmath11 m fields , and faces are plaquette potentials .
this was studied in depth in @xcite .
the classically massless scalars are the little higgs and are associated with non - contractible loops in theory space .
the little higgs has a mass generated radiatively , but the size of the radiative corrections are parametrically smaller than the tev scale . in the limit of vanishing couplings
the nl@xmath11 m fields become exact goldstone bosons under the symmetry @xmath8 .
this leaves the little higgs exactly massless , but also without non - derivative interactions . coupling constants break these chiral symmetries and allow masses for the little higgs to be generated radiatively .
little higgs models always require at least two couplings to communicate sufficient chiral symmetry breaking to generate a mass radiatively , and therefore quadratic divergences first arise at two loop order . to ensure that there are no one loop quadratic divergences to the the little higgs mass there are simple rules for theory space models @xcite .
the first rule is that if all nl@xmath11 m fields are bi - fundamentals , then there are no one loop gauge quadratic divergences .
if no plaquette contains any nl@xmath11 m field more than once , then there one loop quadratic divergences to the little higgs mass from the scalar potential .
having removed the one loop quadratic divergences , there are always finite contributions to the little higgs mass
. we can estimate the size of these finite contributions by studying the theory beneath the scale of the classically massive modes . in this low energy thoery
there are a quadratic divergences and they are cut - off by the mass of the first heavy states and become finite corrections to the mass squared of the little higgs .
the mass of the heavy modes is roughly @xmath12 .
this means that the correction to the little higgs mass from the low energy quadratic divergence is : @xmath13 this is the same size as the two loop quadratic divergence in the full theory .
thus , even if we remove higher order quadratic divergences , we can not push the cut - off higher than two loop factors above the ew scale , setting @xmath14 this still allows for a seperation of scales furthermore , this means that the states that stabilizes the electroweak scale is invisible to roughly @xmath15
the minimal theory space model that has the requisite properties for ewsb has two gauge group and four nl@xmath11 m fields and is illustrated in fig .
[ fig : minimal ] .
the two gauge groups are taken to be @xmath16 and @xmath17 with the @xmath18 embedded as the @xmath19 generator of @xmath16 .
there are two plaquettes added : @xmath20 this gauged nl@xmath11 m breaks the @xmath21 $ ] gauge symmetry down to the diagonal @xmath22 $ ] that is identified as the electroweak gauge group .
the plaquettes give mass to one linear combination of nl@xmath11 m fields and leave three massless .
one is eaten by the massive @xmath16 gauge bosons while two are physical and massless and are the little higgs of the theory . since all the link fields are bi - fundamentals under the two gauge groups and the plaquettes contain each link field only once , there are no one loop quadratic divergences to the little higgs masses .
the scalar potential for little higgs , @xmath23 , is of the form @xmath24 or in terms of the linearized modes : @xmath25 ^ 2 + \cdots\ ] ] where @xmath26 . written in terms of the doublets , the potential is : @xmath27 this is identical to the higgs potential inside the mssm except that the coefficient is undetermined .
hence , the mass of the higgs is undetermined , although we do expect @xmath28 , leading to a physical higgs mass in the range .
the standard model fermions are charged under the @xmath17 gauge group .
it is convenient to write the fermions as triplets under the global @xmath16 , i.e. @xmath29 and @xmath30 .
they are coupled to the little higgs , @xmath31 , through operators such as : @xmath32 this operator induces a one loop quadratically divergent contribution to the little higgs mass . for everything but the top quark ,
the quadratic divergence is negligeable . for the top quark
, the one loop quadratic divergence must be dealt with . by adding an additional massive vector - like fermion , @xmath33 and @xmath34
, we can promote the quark doublet , @xmath35 , to quark triplet under the global @xmath16 , @xmath36 . by preserving the global symmetry of the coupling
, we remove the one loop quadratic divergence to the top quark . the quadratically divergent part of the one loop coleman - weinberg effective potential is : @xmath37 where @xmath38 from the @xmath39 coupling and @xmath40 if without the vector - like fermion @xmath33 and becomes the identity matrix with with @xmath33 . when @xmath41 is the identity matrix
, we preserve the @xmath16 symmetry and the effective potential is independent of @xmath42 .
one loop log divergences from the top sector drives one linear combination of the higgs to acquire a mass and the orthogonal combination are stabilized by one loop log divergences from the gauge and scalar sectors @xcite .
this leads to phenomenologically acceptable ewsb driven radiatively much as in the mssm .
we refer the reader to @xcite for more details .
we have presented new models of ewsb that stabilize the weak scale to radiative corrections and have weakly coupled physics at 1 tev . at the 100 gev scale
it is a 2hdm model with possibly additional scalar singlets and @xmath43 triplets . at 1 tev
there are additional vector bosons that cancel the sms guage @xmath44 divergences , additional scalars that cancel the higgs quartic interaction @xmath44 divergence and additional coloured fermions that cancel the top s @xmath44 divergence .
this description is valid up to @xmath45 tev where the nl@xmath11 m becomes strongly coupled and new physics emerges .
there is a tremendous amount of physics that is still to be done : precision studies @xcite , experimental signature @xcite , exploring other little higgs models @xcite , and uv completing the theory above @xmath45 tev .
n. arkani - hamed , a. g. cohen and h. georgi , phys .
b * 513 * , 232 ( 2001 ) d. b. kaplan and h. georgi , phys .
b * 136 * , 183 ( 1984 ) .
n. arkani - hamed , a. g. cohen , t. gregoire and j. g. wacker , arxiv : hep - ph/0202089 . n. arkani - hamed , a. g. cohen , e. katz , a. e. nelson , t. gregoire and j. g. wacker , arxiv : hep - ph/0206020 . n. arkani - hamed , a. g. cohen and h. georgi , jhep * 0207 * , 020 ( 2002 ) t. gregoire and j. g. wacker , arxiv : hep - ph/0206023
. n. arkani - hamed , a. g. cohen , e. katz and a. e. nelson , jhep * 0207 * , 034 ( 2002 ) i. low , w. skiba and d. smith , arxiv : hep - ph/0207243 .
r. s. chivukula , n. evans and e. h. simmons , arxiv : hep - ph/0204193 .
|
in this note we present a review of the little higgs models that stabilize the electroweak by realizing the standard model higgs as a pseudo - goldstone boson .
|
two types of experimental methods can be used to detect neutrino oscillations : observing the appearance of a neutrino flavour initially absent in the beam or measuring the disappearance rate of the initial flavour . in the latter case , one must know the flux of the beam precisely . in this type of experiment one
explores whether less than the expected number of neutrinos of a produced flavour arrives at a detector or whether the spectral shape changes if observed at various distances from a source .
since the final state is not observed , disappearance experiments can not tell into which flavor a neutrino has oscillated .
an appearance experiment searches for possible new flavours of neutrino , which does not exist in the original beam , or for an enhancement of an existing neutrino flavour .
the identification of the flavour relies on the detection of the corresponding lepton produced in its charged current ( cc ) interactions : @xmath5 with @xmath6 and where x denotes the hadronic final state . in the past two decades ,
several experiments carried out with atmospheric and accelerator neutrinos , as well as with solar and reactor neutrinos , have established the picture of a three - neutrino oscillation scenario with two large mixing angles .
atmospheric sector flavor conversion was first established by the super - kamiokande @xcite and macro @xcite experiments and then confirmed by the k2k @xcite and minos @xcite longbaseline experiments .
the chooz @xcite and palo verde @xcite reactor experiments excluded indirectly the @xmath7 channel as the dominant process in the atmospheric sector .
however , the direct observation of flavour transition through the detection of the corresponding lepton has never been observed .
appearance of @xmath4 will prove unambiguously that @xmath8 oscillation is the dominant transition channel at the atmospheric scale .
the opera experiment @xcite has been designed to directly observe the appearance of @xmath4 in a pure @xmath3 beam on an event by event basis .
the @xmath4 signature is given by the decay topology and kinematics of the short lived @xmath9 leptons produced in the interaction of @xmath10 and decaying to one prong ( @xmath11 or @xmath12 ) or three prongs , which are @xcite : @xmath13
the cngs @xmath3 beam produced by the cern - sps is directed towards the opera detector , located in the gran sasso underground laboratory ( lngs ) @xcite in italy , 730 km away from the neutrino source at cern . in order to study @xmath14 oscillations in appearance mode
as indicated in the atmospheric neutrino sector , the cern neutrinos to gransasso ( cngs ) neutrino beam @xcite was designed and optimized by maximizing the number of @xmath4cc interactions at the lngs .
the average @xmath3 beam energy is 17 @xmath15 , well above tau production energy treshold .
the @xmath16 contamination is @xmath17 in flux , @xmath18 in terms of interactions .
the @xmath19 and @xmath20 contaminations are lower than @xmath21 , while the number of prompt @xmath4 from @xmath22 decay is negligible .
the average @xmath23 ratio is 43 @xmath24 , suitable for oscillation studies at atmospheric @xmath25 . due to the earth s curvature neutrinos from cern
enter the lngs halls with an angle of about @xmath26 with respect to the horizontal plane . with a nominal cngs beam intensity of @xmath27 protons on target ( @xmath2 ) per year , and assuming @xmath28 and full mixing , about 10 @xmath4 events are expected to be observed in opera in 5 years of data taking , with selection criteria reducing the background to 0.75 events .
the goal is to accumulate a statistics of neutrino interactions correspomding to @xmath29 @xmath2 in 5 years .
the 2008 , 2009 and 2010 runs achieved a total intensity of @xmath30 , @xmath31 and @xmath32 @xmath2 respectively . within these three years , neutrinos produced 9637 beam events .
the processing of these events , particularly the scanning of emulsion films , is continuously going on .
the 2011 run started on may 2011 and is still in progress .
at the cngs energies the average @xmath9 decay length is submillimetric , so opera uses nuclear emulsion films as high precision tracking device in order to be able to detect such short decays .
emulsion films are interspaced with @xmath33 thick lead plates , which act as neutrino target and form the largest part of the detector mass
. this technique is called emulsion cloud chamber ( ecc ) .
it was successfully used to establish the first evidence for charm in cosmic rays interactions @xcite and in the donut experiment @xcite for the first direct observation of the @xmath4 .
to date , nine @xmath4 cc interactions have been observed by donut produced by a fixed target 800 @xmath15 proton beam configuration .
opera is a hybrid detector made of two identical super modules ( sm1 and sm2 ) , each one formed by a target section and a muon spectrometer as shown in figure [ fig : detector ] .
each target section is organized in 31 vertical `` @xmath34 '' , transverse to the beam direction .
walls are filled with `` @xmath35 @xmath36 '' with an overall mass of 1.25 @xmath37 .
they are followed by double layers of scintillator planes acting as target trackers ( tt ) that are used to locate neutrino interactions occurred within the target .
a target brick consists of 56 lead plates of 1 mm thickness interleaved with 57 emulsion films .
the lead plates serve as neutrino interaction target and the emulsion films as 3-dimensional tracking detectors providing track coordinates with a sub - micron accuracy and track angles with a few mrad accuracy .
the material of a brick along the beam direction corresponds to about 10 radiation length and 0.33 interaction length .
the brick size is @xmath38 and its weight is about @xmath39 . in order to reduce the emulsion scanning load , changeable sheets ( cs )
@xcite film interfaces have been used .
they consist in tightly packed doublets of emulsion films glued to the downstream face of each brick .
charged particles from a neutrino interaction in a brick cross the cs and produce signals in the tt that allow the corresponding brick to be identified and extracted by an automated brick manipulator system ( bms ) .
the spectrometers consist of a dipolar magnet instrumented with active detectors , planes of rpcs ( internal tracker , it ) and drift tubes ( precision tracker , pt ) .
tasks of the spectrometers are muon identification and charge measurement in order to minimize the background . for muon momenta between 2.5 @xmath40 and 45 @xmath40 ,
the fraction of events with wrong charge determination is @xmath41 .
the @xmath42 to @xmath43 events ratio , within the selected momentum range , obtained from data can be directly compared with predictions based on monte carlo simulations : @xmath44 for data , @xmath45 for mc .
figure [ fig : mom ] left - side shows the momentum and momentum times charge distribution for data and mc .
charge ) : data ( black dots with error bars ) and mc ( solid line ) are normalised to one .
left : bjorken - y variable reconstructed in data ( dots with error bars ) and mc ( shaded areas ) .
the mc distributions are normalised to data .
the different contributions of the mc are shown in different colours : qe + res contribution in light grey , dis contribution in grey and the nc contamination in dark grey.,title="fig : " ] charge ) : data ( black dots with error bars ) and mc ( solid line ) are normalised to one . left : bjorken - y variable reconstructed in data ( dots with error bars ) and mc ( shaded areas ) .
the mc distributions are normalised to data .
the different contributions of the mc are shown in different colours : qe + res contribution in light grey , dis contribution in grey and the nc contamination in dark grey.,title="fig : " ] in figure [ fig : mom ] right - side , bjorken - y distribution is shown for the events with at least a muon track .
the agreement between data and mc simulation is reasonable .
the sum of the qe and res processes can be clearly seen as a peak at low y values .
the nc contribution shows up at values of bjorken - y close to one .
the nc contribution becomes negligible when a track with its momentum measured by the spectrometer is required .
a detailed description of the complete detector can be found in @xcite .
event reconstruction procedures and a performances of the opera electronic detectors can be found in more detail in @xcite .
neutrino event analysis starts with the pattern recognition in the electronic detectors . charged particle tracks produced in a neutrino interaction
generate signals in the tt and in the muon spectrometer .
a brick finding algorithm is applied in order to select the brick which has the maximum probability to contain the neutrino interaction .
the brick with the highest probability is extracted from the detector for analysis .
the efficiency of this procedure reaches @xmath46 in a subsample where up to 4 bricks per event were processed .
after extraction of the brick predicted by the electronic detectors , its validation comes from the analysis of the cs films .
the measurement of emulsion films is performed through high - speed automated microscopes @xcite with a sub - micrometric position resolution and angular resolution of the order of one milliradian . if no expected charged track related to the event is found in the cs , the brick is returned back to the detector with another cs doublet attached . if any track originating from the interaction is detected in the cs , the brick is exposed to cosmic rays ( for alignment purposes ) and then depacked .
the emulsion films are developed and sent to the scanning laboratories of the collaboration for event location studies and decay search analysis .
all the track information of the cs is then used for a precise prediction of the tracks in the most downstream films of the brick ( with an accuracy of about @xmath47 ) . when found in this films ,
tracks are followed upstream from film to film .
the scan - back procedure is stopped when no track candidate is found in three consecutive films and the lead plate just upstream the last detected track segment is defined as the vertex plate . in order to study the located vertices and reconstruct the events , a general scanning volume is defined with a transverse area of @xmath48 for 5 films upstream and 10 films downstream of the stopping point . all track segments in this volume are collected and analysed .
after rejection of the passing through tracks related to cosmic rays and of the tracks due to low energy particles , the tracks produced by the neutrino interaction can be selected and reconstructed .
the present overall location efficiency averaged over nc and cc events , from the electronic detector predictions down to the vertex confirmation , is about @xmath49 .
once the neutrino interaction is located , a decay search procedure is applied to detect possible decay or interaction topologies on tracks attached to the primary vertex .
the main signature of a secondary vertex ( decay or nuclear inetaraction ) is the observation of a track with a significant impact parameter ( ip ) relative to the neutrino interaction vertex .
the ip of primary tracks is smaller than 10@xmath50 after excluding tracks produced by low momentum particles .
when secondary vertices are found in the event , a kinematical analysis is performed , using particle angles and momenta measured in the emulsion films . for charged particles up to about 6 @xmath40 , momenta can be determined using the angular deviations produced by multiple coulomb scattering ( mcs ) of tracks in the lead plates @xcite with a resolution better than @xmath51 . for higher momentum particles ,
the measurement is based on the position deviations .
the resolution is better than @xmath52 on 1/p up to 12 @xmath40 for particles passing through an entire brick .
a @xmath53-ray search is performed in the whole scanned volume by checking all tracks having an ip with respect to the primary or secondary vertices lower than 800@xmath50 .
the angular acceptance is @xmath54 @xmath55 .
the @xmath53-ray energy is estimated by a neural network algorithm that uses the number of segments , the shape of the electromagnetic shower and also the mcs of the leading tracks .
in the following , the analysis results @xcite of about @xmath56 of the 2008 and 2009 data sample , corresponding to the @xmath57 @xmath58 are presented . the decay search procedure was applied to a sample of 1088 events of which 901 were classified as cc interactions . in the sample of cc interactions ,
20 charm decay candidates were observed , in good agreement with the expectations from the monte carlo simulation , @xmath59 . out of them 3 have a 1-prong topology where @xmath60 was expected .
the background for the total charm sample is about 2 events .
several @xmath19-induced events have also been observed .
moreover , a first cc @xmath4 candidate has been detected .
the expected number of @xmath4 events detected in the analysed sample is about @xmath61 at @xmath62 @xmath63 and full mixing .
in this section , the first tau neutrino candidate @xcite will be described .
the location and decay search procedure yielded a neutrino interaction vertex with 7 tracks .
one track exhibits a visible kink with an angular change of @xmath64 after a path length of @xmath65 .
the kink daughter momentum is estimated to be @xmath66 @xmath40 by mcs measurement and its transverse momentum to the parent direction is @xmath67 @xmath68 .
the event is displayed in figures [ fig : tau ] and [ fig : tau2 ] .
candidate event.,title="fig : " ] candidate event.,title="fig : " ] candidate event . ]
all the tracks from the neutrino interaction vertex were followed until they stop or interact .
the probability that one of them is left by a muon is estimated to be less than @xmath69 .
the residual probability for being a @xmath3cc event , with a possibly undetected large angle @xmath70 track , is about @xmath21 ; a nominal value of @xmath71 is assumed .
none of the tracks is compatible with being an electron .
two electromagnetic showers caused by @xmath53-rays , associated with the event , have been located and studied . the energy of @xmath72 is @xmath73 @xmath15 and it is clearly pointing to the decay vertex .
the @xmath74 has an energy of @xmath75 @xmath15 and it is compatible with pointing to either vertex , with a significantly larger probability to the decay vertex .
all the selection cuts used in the analysis were those described in detail in the experiment proposal @xcite and its addendum @xcite .
all the kinematical variables of the event and the cut applied are given in table [ tab : kinematic ] .
.kinematical variables of @xmath4 candidate event .
[ cols="<,^,<",options="header " , ] the invariant mass of the two observed @xmath53-rays is @xmath76 supporting the hypothesis that they are emitted in a @xmath77 decay . the invariant mass of the charged decay daughter assumed to be a @xmath78 and of the two @xmath53-rays amount to @xmath79 @xmath68 , which is compatible with the @xmath80 mass .
so the decay mode of the candidate is consistent with the hypothesis @xmath81 ( where the branching ratio is about @xmath82 ) .
the two main sources of background to the @xmath83 channel where a similar final state may be produced are : * the decays of charmed particles produced in @xmath3 cc interactions where the primary muon is not identified as well as the @xmath84 pair production in @xmath3 nc interactions where one charm particle is not identified and the other decays to a 1-prong hadron channel ; * the 1-prong inelastic interactions of primary hadrons produced in @xmath3cc interactions where the primary muon is not identified or in @xmath3 nc interactions and in which no nuclear fragment can be associated with the secondary interaction .
the monte carlo expectation of the first background source is @xmath85 event , the fraction produced in @xmath86 cc interactions is less than @xmath69 events , the second type of background amounts to @xmath87 event .
the total background in the decay channel to a single charged hadron is @xmath88 events .
the probability that this background events fluctuate to one event is @xmath89 @xmath90 . as the search for @xmath9 decays
is extended to all four channels , the total background then becomes @xmath91 .
the probability that this expected background to all searched decay channels of the @xmath9 fluctuates to one event is @xmath92 @xmath93 . at @xmath94 @xmath63 and full mixing ,
the expected number of observed @xmath9 events with the present analyzed statistics is @xmath95 of which @xmath96 in the one - prong hadron topology , compatible with the observation of one event .
during 2008 , 2009 and 2010 runs , a total intensity of @xmath30 , @xmath31 and @xmath32 @xmath2 respectively , was achieved . within these three years
, 9637 beam events have been collected within the opera target .
the neutrino interaction location and decay search are going on .
a first candidate @xmath4 cc interaction in the opera detector at lngs was detected after analysis of a sample of events corresponding to @xmath57 @xmath2 in the cern cngs @xmath3 beam .
the expected number of @xmath4 events in the analysed sample is @xmath97 .
the candidate event passes all selection criteria , it is assumed to be a @xmath9 lepton decaying into @xmath98 .
the observation of one possible tau candidate in the decay channel @xmath99 has a significance of @xmath100 of not being a background fluctuation .
http://proj - cngs.web.cern.ch / prj - cngs/. . ; . .
; + ; + . ; + phd .
thesis , universite de savoie , 2008 , lapp - t-2008 - 02 . .
cern - spsc-2000 - 028 ; lngs p25/2000 .
cern - spsc-2001 - 025 ; lngs - exp 30/2001 add .
|
the opera experiment in the underground gran sasso laboratory ( lngs ) has been designed to perform the first detection of neutrino oscillations in direct appearance mode in the muon neutrino to tau neutrino channel .
the detector is hybrid , being made of an emulsion / lead target and of electronic detectors .
it is placed in the cngs neutrino beam 730 @xmath0 away from the neutrino source .
runs with cngs neutrinos were successfully carried out in 2008 , 2009 , and 2010 . after a brief description of the beam and the experimental setup
, we report on event analysis of a sample of events corresponding to @xmath1 @xmath2 in the cern cngs @xmath3 beam that yielded the observation of a first candidate @xmath4 cc interaction .
the topology and kinematics of this candidate event are described in detail .
the background sources are explained and the significance of the candidate is assessed .
[ 1999/12/01 v1.4c il nuovo cimento ]
|
quality assessment of photographs is not a new issue arising with digital cameras .
birkhoff , back in 1933 proposed that the aesthetic appeal of objects relates to the ratio of order and complexity in images .
the hardness in this task is to define order and complexity [ 4 ] . with the advent of digital cameras and smartphones ,
people capture more photographs than they can consume . nowadays
, social media provide an adequate filter of our social graph , displaying to us photographs and posts by our friends or friends of friends , based on popularity metrics .
what if we could see photographs based on our taste , previous likes , or general aesthetic criteria ? in the field of information retrieval , the content based image retrieval ( cbir ) systems apply computer vision techniques to the image retrieval problem , by searching in large databases . in this context
, we do not search for metadata or documents , but we decompose the image on its pixels , attempting to draw inferences between its content and its aesthetic value .
this relatively recent research area is termed as `` computational aesthetics '' .
researchers in this area leverage techniques from image processing and computer vision , combining methodologies from psychology and art theory .
[ 32,22 ] the field of computational aesthetics gradually attracts interest in the scientific community .
scopus data ( figure 1 ) reveal the steady growth in literature regarding this area . in 2014 , 89 papers were published in international conferences and journals . if we dig more on scopus data , we find that the majority of publications comes from asia ( national university of singapore , universiti tenaga nasional and zhejiang university ) and north america ( simon fraser university , carnegie mellon university , and georgia institute of technology )
. coordinates ( 1990 , 1 ) ( 1991 , 2 ) ( 1992 , 1 ) ( 1993 , 2 ) ( 1994 , 2 ) ( 1995 , 5 ) ( 1996 , 13 ) ( 1997 , 7 ) ( 1998 , 9 ) ( 1999 , 8) ( 2000 , 7 ) ( 2001 , 11 ) ( 2002 , 27 ) ( 2003 , 19 ) ( 2004 , 16 ) ( 2005 , 29 ) ( 2006 , 34 ) ( 2007 , 33 ) ( 2008 , 43 ) ( 2009 , 37 ) ( 2010 , 56 ) ( 2011 , 45 ) ( 2012 , 71 ) ( 2013 , 74 ) ( 2014 , 89 ) ( 2015 , 67 ) ; researchers involved in aesthetic assessment of photographs try to correlate low or high - level attributes of an image with its quality evaluation .
this process , understandably contains subjectivity in gathering ground truth data .
the solution proposed is to combine these extracted features with crowdsourced ratings of photographs , ideally of big populations [ 30 ] .
but , before we can design features to assess quality , we should decide on the perceptual criteria that we use when we judge photographs .
literature groups three factors on judging the quality of photographs : simplicity , realism and basic photography techniques [ 15 ] .
* simplicity*. one distinctive aspect of professional shots to everyday snapshots is that , pro photos are simple and easy to separate the foreground from the background . the tricks used by professionals in order to distinguish the background from the subject is by widening the lens aperture , increasing the natural color contrast of the subject and the lighting contrast .
* realism*. our snapshots usually look ordinary and quite common , while the professional shots look surreal . to achieve this effect , professionals are aware of the lighting conditions of dusk or dawn , choosing to shoot at hours that the sunlights are indirect . also , they use hardware filters to make the sky bluer ( nd filters ) , adjusting the color palette and saturation levels .
hardware in general , is more expensive and customized than regular smartphones , allowing to tweak the aperture , shutter speed , lens optical zoom and lots of other options . by leveraging the above techniques ,
professionals tend to shoot more unusual objects and situations .
* basic photography techniques*. it is very rare for a professional photo to be entirely blurry .
blurriness is often the result of camera shake or low quality lens .
background blurriness , though , could be an aspect of professional deliberate result , as described above .
in addition , professionals shoot usually higher contrast photos than point - and - shoot snapshot users .
given the above grouping , researchers try to model these factors into features that can be extracted from images .
we will describe those features in section 2 .
the rest of the paper is structured as follows : in section 3 we describe the aesthetic benchmark datasets used in computational evaluation , in section 4 we asses the machine learning frameworks of predicting aesthetic value and , finally , in section 5 we discuss the limitations and future work .
in the last fifteen years there have been significant contributions to the field of image representation and feature extraction towards semantic understanding . [ 14 ] aesthetics and emotional value are based on semantics , so it is not surprising that they draw inspiration from the above fields . in the literature
we notice a spectrum of low level ( edges , textures , color histograms etc . ) and high level ( rule of thirds , symmetry , saliency , face recognition etc . ) features combined .
we group all the available features mentioned in literature as follows : * _ color , texture , composition , content_*. also , we attempt a qualitative labelling of each feature according to its implementation complexity with current image processing tools and frameworks .
note that the implementation complexity might overlap with computation complexity particularly in data intensive extraction techniques , such as facial recognition .
* hsl and hsv color spaces*. hue , saturation and lightness ( hsl ) and hue , saturation and value ( hsv ) are the two most common cylindrical - coordinate representations of points in an rgb color model .
they are used in literature due to the better representation of color than rgb .
features often extract the average value of each channel in the whole image , or segments of it
. * white balance .
* professional cameras have to take into account the `` color temperature '' of the light source .
on the other hand , cheap cameras tend to shoot blue - ish photographs due to low - quality lenses .
we estimate the average color temperature distribution . *
contrast*. one of the most important aspects of a photograph is the difference in colour luminance that makes an object or thing distinguishable .
we create a multi - scale contrast map from the brightness histogram of the photo and count average and median values .
* pixel intensity*. too much exposure often creates lower quality pictures .
those that are too dark are often also not appealing .
thus light exposure can often be a good discriminant between high and low quality photographs .
however , an over / under - exposed photograph under certain scenarios may yield very original and beautiful shots .
[ 8 ] we use the average pixel intensity to estimate the use of light , where iv is the value channel of hsv from each row x , column y of the image .
@xmath0 * pleasure , arousal , dominance*. emotional coordinates based on saturation and brightness are estimated by [ 32 ] from their psychology experiments .
this relationship between the saturation and brightness is modelled as follows : @xmath1 @xmath2 @xmath3 where v , s are the value and saturation metrics of the hsv color model .
* color templates*. there are several metrics proposed by [ 18 ] in order to estimate the distance between the hue distribution and a certain color template . in particular , munsel color system provides efficient regions in a color wheel according to its position . in ( fig 2 )
we see the hue distribution models , where the gray color indicates the efficient regions that result in harmony . [ 35 ] proposed an improved metric by approximating the color distribution of hue histogram by identifying its location of peak value .
the distance between the hue histogram distribution and the most matching munsel color template is expressed as : @xmath4 @xmath5 where k=1 ... 7 is the type of color template ( see above fig 2 ) , dist(i , rk ) is the distance between hue value i and the gray region rk of template tk , h(i ) stands for the occurrence of hue value i in the histogram , dist(h(i ) , rk ) is zero when i is in the gray regions rk arc = length distance between i and the nearest border of gray regions and s(i ) is the average saturation of hue value i and acts as weight on colors with low saturation .
then , we should find which template fits better our test image .
ideally this is the color template with the minimum distance .
however some templates contain more gray than others , so there must be some bias .
we overcome this by measuring the fitness of color templates @xmath6 where nk is the total number of pixels in gray regions , is the image size and pk the inverse proportion of gray region size in the template . combining the ratio rk and distance dk
, we measure the similarity between hue histogram and color templates : @xmath7 * color moment*. a group of standard measures which characterise color distribution in an image in the same way that central moments describe a probability distribution uniquely .
we measure mean , standard deviation , skewness and kurtosis , for each hsv channel .
* colourfulness*. [ 8 ] proposed a robust method to find the relative color distribution , distinguishing colorful images from monochromatic and low contrast ones .
they employed the earth mover s distance ( emd ) which is a measure of similarity between any two weighted distributions . dividing the colour space into n cubic blocks with four equal partitions along each dimension , they take each such cube as a sample point .
distribution d1 is generated as the color distribution of a hypothetical image such that for each of n sample points , the frequency is 1/n .
distribution d2 is computed from the test image , by finding the frequency of occurrence of color within each of the n cubes .
but , emd requires the pairwise distance between the two under - comparison points , so we count the pairwise euclidean distances between the geometric centers c of each cube . finally the colourfulness measure is estimated as follows : @xmath8 * colour names*. we consider that every pixel of an image can be assigned to one of the following groups : black , blue , brown , green , gray , orange , pink , purple , red , white , yellow .
the algorithm proposed by [ 33 ] mimics the way humans judge a whole photo of an image by its color .
this measure let us decide on the style of the photographer as well .
.color features [ cols="<,<,<,<",options="header " , ] * image and processing complexity*. these features are inspired by birkhoff s idea that aesthetic appeal of objects relates to the ratio of order and complexity .
the optimal image is accomplished by an image that has a low image complexity and a low processing complexity , at the same time .
[ 27,28,21 ] the idea states that compression error correlates with complexity .
they employ shannon s entropy in order to measure the image complexity .
the processing complexity is estimated via the kolmogorov complexity .
* visual weight ratio*. [ 12 ] proposed a method to recover the surface layout from an outdoor image using geometric context .
the scene is segmented into sky regions , ground regions , and vertical standing objects , using adaboost on a variety of low level image features .
we take vertical standing objects as subject areas . that way we distinguish between clear , cloudy and sunset skies .
the ratios between the areas of these foreground - background rectangles should be close to the golden ratio for a better appeal .
* foreground position*. [ 3 ] modelled the relative foreground position as the normalized euclidean distance between the foreground s center of mass ( visual attention center ) , to each of four symmetric stress points ( rule of thirds intersections ) in the image frame . this
metric is ideal for single - subject scenes such as animals or portraits but might me ineffective in landscapes or seaside view which do not have a clear foreground . for these situations
we choose the above technique of visual weight ratio .
* salient regions*. saliency of a subject is the quality by which it stands out compared to its neighbors .
[ 37 ] extract the salient regions from an image by utilizing a visual saliency model .
we assume that the salient regions contain the photo subject .
we first find the salient locations in the original image , compute the saliency map , the segmented image , and end up with the salient mask based on the main salient locations .
our target is to enhance those salient regions as foreground objects .
[ 38 ] this saliency retargeting problem ( fig 3 ) is solved by sequential quadratic programming ( sqp ) .
* graphlets*. there are usually many components within a photo . among these components , a few spatially neighboring ones and their interactions capture photo local aesthetics . since a graph is a powerful tool to describe the relationships between objects , [ 39 ] used graphs to model the spatial interactions between image components .
their technique is to segment a photo into a set of atomic regions using unsupervised fuzzy clustering .
based on this , they extract graphlets graphlet is a small - sized connected graph defined as : g = ( v , e ) where v is a set of vertices representing locally distributed atomic regions and e is a set of edges , each of which connects pairwise spatially adjacent atomic regions .
then , the graphlets are projected onto a manifold and the authors subsequently proposed an embedding algorithm .
* pyramid of histograms of orientation gradients*. [ 26 ] proposed a model based on phog following a pyramid approach to calculate the hog .
hog values are calculated based on the maximum gradient magnitudes in the color channels .
based on the new gradient image , we estimate the high self - similarity , moderate complexity and low anisotropy .
* symmetry*. [ 30 ] measured the symmetry of an image based on the difference of the histogram of oriented gradients ( hog ) [ 7 ] between the first half of the image and its opposite right half . *
faces*. recognition of active faces inside an image is still an open research field ; the researchers of computational aesthetics borrow the advancements of this field in order to count the number of frontal faces , the relative size of the biggest face as well as the shadow area in faces . [ 20 ] the implementation used is usually the state of art algorithm proposed by [ 34 ] .
* skin*. along with faces , skin color and recognition is important in order to identify a photograph containing humans .
we estimate the number of skin pixels and the relative amount of skin with respect to the size of faces . the key concept is to detect the color space of pink that corresponds to human skin [ 19 ] .
the above features are extracted usually from ground truth aesthetic images found in various dataset created for this task .
researchers test their techniques with the above benchmark datasets : dpchallenge , photo.net , flickr , terragalleria , alipr and ava .
[ 14 ] * * dpchallenge * * allows users to participate and contest in theme - based photography on diverse themes such as life and death , portraits , animals , geology , street photography .
peer rating on overall quality , on a 1- 10 scale , determines the contest winners .
( 16,509 images ) * * photo.net * * is a platform for photography enthusiasts to share and have their pictures peer rated on a 1 - 7 scale of aesthetics .
the photography community also provides discussion forums , reviews on photos and photography products , and galleries for members and casual surfers .
( 20,278 images ) * * flickr * * is one of the most popular online photo - sharing sites in the world . according to flickr , the interestingness of a picture is dynamic and depends on a plethora of criteria including its photographer , who marks it as a favorite , comments , and tags given by the community . * * terragalleria * * displays travel photography of quang - tuan luong ( a scientist and a photographer ) , and is one of the finest resources for u.s . national park photography on the web .
all photographs here have been taken by one person ( unlike other benchmarks ) , but multiple users have rated them on overall quality on a 1 - 10 scale . *
* alipr * * is a web - based image search and tagging system that also allows users to rate photographs along 10 different emotional categories such as surprising , amusing and adorable . *
* ava * * is the biggest aesthetics dataset which contains a rich variety of metadata including a large number of aesthetic scores for each image , semantic labels for over 60 categories as well as labels related to photographic style .
( 250,000 images )
combining the above ground truth aesthetic datasets with the features extracted of their images , researchers employ machine learning techniques in order to predict the aesthetic value of an image . in this process
, both classification and regression is performed for different desired results . with classification
we can distinguish between an aesthetic photo or not .
but this measure tends to be kind of absolute in such a subjective task .
however , aesthetics is a more abstract quality and quantifying it requires a relatively larger numeric scale .
thus , regression techniques provides us with a scale of aesthetic score for each photo , so we are able to compare why a photo is considered of better quality from another .
svm , svr and classification and regression trees ( cart ) are the most usual methods used for classification and regression .
in addition , forms of unsupervised learning like the k - means clustering are used for visual vocabulary generation and graph - based region segmentation , the state of the art achieved on the above datasets varies according with the implementation and the training dataset size , making it difficult to compare the accuracy results of the proposed techniques .
the best reported accuracies in most cases is around 80% [ 8,35,39 ] .
the field of computational aesthetics seems promising considering the growing interest from the research community .
researchers try to model the aesthetic quality of photographs into low and high - level features , drawing inspiration from art theory , psychology and marketing .
there are many quite imaginative applications of computational aesthetics such as coral reef evaluation image aesthetics [ 10 ] , optimal text placement inside an image [ 18 ] , better route suggestion in a city based on surroundings [ 25 ] or personality prediction [ 29 ] some underexplored areas proposed by this review are : personalization feedback in predictions , non - photographic images , black and white artistic consideration , and exif camera metadata ..
99 battiato , s. , moltisanti , m. , ravi , f. , bruna , a. r. , naccari , f. ( 2013 , february ) .
aesthetic scoring of digital portraits for consumer applications . in ist / spie electronic imaging ( pp .
866008 - 866008 ) .
international society for optics and photonics .
bhattacharya , s. , sukthankar , r. , shah , m. ( 2010 , october ) . a framework for photo - quality assessment and enhancement based on visual aesthetics . in proceedings of the 18th acm international conference on multimedia ( pp .
271 - 280 ) .
acm .
dalal , n. , triggs , b. ( 2005 , june ) .
histograms of oriented gradients for human detection . in computer vision and pattern recognition , 2005 .
cvpr 2005 .
ieee computer society conference on ( vol .
886 - 893 ) .
ieee .
dhar , s. , ordonez , v. , berg , t. l. ( 2011 , june ) .
high level describable attributes for predicting aesthetics and interestingness . in computer vision and pattern recognition ( cvpr ) , 2011 ieee conference on ( pp .
1657 - 1664 ) .
ieee .
ke , y. , tang , x. , jing , f. ( 2006 , june ) .
the design of high - level features for photo quality assessment . in computer vision and pattern recognition , 2006 ieee computer society conference on ( vol .
419 - 426 ) .
lai , c. y. , chen , p. h. , shih , s. w. , liu , y. , hong , j. s. ( 2010 ) .
computational models and experimental investigations of effects of balance and symmetry on the aesthetics of text - overlaid images . international journal of human - computer studies , 68(1 ) , 41 - 56 .
li , c. , loui , a. c. , chen , t. ( 2010 , october ) . towards aesthetics : a photo quality assessment and photo selection system . in proceedings of the 18th acm international conference on multimedia ( pp .
827 - 830 ) .
acm .
machajdik , j. , hanbury , a. ( 2010 , october ) .
affective image classification using features inspired by psychology and art theory . in proceedings of the international conference on multimedia ( pp .
83 - 92 ) .
nishiyama , m. , okabe , t. , sato , i. , sato , y. ( 2011 , june ) .
aesthetic quality classification of photographs based on color harmony . in computer vision and pattern recognition ( cvpr ) , 2011 ieee conference ( pp .
33 - 40 ) .
ieee .
quercia , d. , ohare , n. k. , cramer , h. ( 2014 , february ) .
aesthetic capital : what makes london look beautiful , quiet , and happy ? . in proceedings of the 17th acm conference on computer supported cooperative work social computing ( pp .
945 - 955 ) .
redies , c. , amirshahi , s. a. , koch , m. , denzler , j. ( 2012 , october ) .
phog - derived aesthetic measures applied to color photographs of artworks , natural scenes and objects . in computer vision
eccv 2012 .
workshops and demonstrations ( pp . 522 - 531 ) .
springer berlin heidelberg .
romero , j. , machado , p. , carballal , a. , osorio , o. ( 2011 ) . aesthetic classification and sorting based on image compression . in applications of evolutionary computation ( pp .
394 - 403 ) .
springer berlin heidelberg .
segalin , c. ; perina , a. ; cristani , m. ; vinciarelli , a. , ( 2016 ) the pictures we like are our image : continuous mapping of favorite pictures into self - assessed and attributed personality traits , in affective computing , ieee transactions on , vol.pp , no.99 , pp.1 - 1 schifanella , r. , redi , m. , aiello , l. m. an image is worth more than a thousand favorites : surfacing the hidden beauty of flickr pictures . in icwsm15 : proceedings of the 9th aaai international conference on weblogs and social media .
aaai wei - ning , w. , ying - lin , y. , sheng - ming , j. ( 2006 , october ) .
image retrieval by emotional semantics : a study of emotional space and feature extraction . in systems ,
man and cybernetics , 2006 .
ieee international conference on ( vol .
3534 - 3539 ) .
ieee .
zhang , l. , gao , y. , zimmermann , r. , tian , q. , li , x. ( 2014 ) .
fusion of multichannel local and global structural cues for photo aesthetics evaluation.image processing , ieee transactions on , 23(3 ) , 1419 - 1429 .
|
researchers try to model the aesthetic quality of photographs into low and high - level features , drawing inspiration from art theory , psychology and marketing .
we attempt to describe every feature extraction measure employed in the above process .
the contribution of this literature review is the taxonomy of each feature by its implementation complexity , considering real - world applications and integration in mobile apps and digital cameras . also , we discuss the machine learning results along with some unexplored research areas as future work . image processing , feature extraction , photograph quality assessment
|
in recent years , the discovery of accelerating expansion of the universe is an amazing development .
it was firstly discovered by observing type ia supernova ( sne ia ) @xcite , which can be used as standard candles @xcite . the cosmic microwave background ( cmb )
measurements from wilkinson microwave anisotropy probe ( wmap ) @xcite and the large scale structure survey by sloan digital sky survey ( sdss ) @xcite confirm this accelerating expansion universe model .
there are two kinds of ideas , i.e. the existence of the dark energy or modifications of the gravity theory , to explain this concept .
the first scheme is most popularly discussed , and many models have been proposed , such as the holographic dark energy models @xcite and the chaplygin gas @xcite .
in addition there are also many modified gravity models , such as the brane world @xcite and @xmath9 @xcite and so on .
the dark energy equation of state(eos ) which is the ratio of pressure to energy density , is a prefect quantity to study the behavior of dark energy .
if eos is a constant @xmath10 , dark energy is the @xmath11 cold dark matter model(lcdm ) , and maybe dark energy is vacuum energy .
otherwise the dark energy would be dynamical scalar field , such as the quintessence @xcite , the quintom@xcite .
there are several way to explore the behavior of dark energy eos .
the most popular way is to build a functional form for eos in terms of some free parameters .
lots of eos parameterizations have already been discussed in the literature(such as @xcite and refs . therein ) .
another way is picking a simple local basis representation for @xmath12 ( bins , wavelets ) , and estimate the associated coefficients @xcite .
in addition , there are also some nonparametric way @xcite . in this paper , we consider a new parameterization for the dark energy eos .
in@xcite , the author developed a new dark energy eos parameterization : @xmath13 .
it is obvious that when @xmath14 , @xmath12 has poor behavior and becomes infinite .
this dark energy eos parameterization can only describe the behavior of dark energy when @xmath2 is not very large . to avoid this problem
, we consider a new dark energy eos parameterization , which is the modification of efstathiou parameterization : @xmath15 .
the value of @xmath12 is @xmath3 at present and @xmath12 becomes to ( @xmath16 ) when @xmath14 . in this model
there are three parameters in all , which is @xmath3 , @xmath4 and @xmath17 . as shown in@xcite
, this eos will get to a nonphysical value in the far future time when redshift @xmath2 approaches @xmath10 , namely , @xmath18 will grow rapidly and diverge . in this paper
, we perform a global data fitting analysis on this new dark energy eos parametrizations , and present constraints on the model parameters from the current observational data , including the seven - year wmap data , baryon acoustic oscillations ( bao ) data , observational hubble data and sn union2 sample .
since dark energy parameters are tightly correlated to some other cosmological parameters , such as the matter density parameter @xmath17 and the hubble constant @xmath19 , it is necessary to consider a global fit procedure in the investigation of the dynamical dark energy .
the paper is organized as follows : in section ii , we review the new dark energy eos parameterization . in section iii , we describe current observational data we used . in section iv , we perform the cosmic observation constraint , the results are also presented .
the last section is the conclusion .
let us start by presenting some of the most investigated eos parameterizations(see also @xcite for other parameterizations ) : @xmath20 where @xmath3 is the current value of the eos , and @xmath4 indicate the revolution of the dark energy eos .
the value of these parameter is determined by the observational data . when @xmath21 and @xmath22 , the dark energy model becomes to lcdm model .
the first parameterization represents a good fit for low redshifts , but has serious problems to explain high-@xmath2 observations since it blows up as @xmath23 when @xmath24 and @xmath25 . for example
, it can not explain the estimated ages of high-@xmath2 objects @xcite .
the second one solves this problem , since @xmath0 is a well behaved function for @xmath1 and recovers the linear behavior in @xmath2 at low redshifts .
the latter was introduced by efstathiou @xcite .
it was built empirically to adjust some quintessence models at @xmath26 .
when @xmath2 approachs infinity , @xmath12 has poor behavior and becomes infinite and this is unnatural .
similar to the second model , let us consider the following eos parameterization [ eqn::eos2 ] w(z ) = w_0+w_a(1 + ) . where the value of @xmath12 is @xmath3 at present and @xmath27 when @xmath14 .
if there is no interaction between dark energy and other component of the universe , one can show from the energy conservation law @xmath28 $ ] that the dark energy density evolves as @xmath29 \frac{dz'}{1+z ' } \right ) , \label{eq : rho phi ( z)}\ ] ] where @xmath30 is the critical density , it is defined by the following equation @xmath31 in this new parameterization , it is hard to write out the analysis formula of this quantity , it is calculated through numerical method .
, @xmath32 ; green line : @xmath33 , @xmath34 ; black line : @xmath35 , @xmath36 . ] in order to study the evolution of cosmological perturbations , we use the public parameterized post - friedmann ( ppf ) package developed by wayne hu(see e.g. @xcite for detailed discussions on ppf method ) to calculate the cmb anisotropy spectrum .
in figure [ fig : cls ] we plot the anisotropy spectrum for different choices of @xmath3 and @xmath4 .
we observe some obvious differences with respect to the lcdm case on larger scales ( multipole number @xmath37 ) . and for the model @xmath38 , there is a slit deviation on the peaks of anisotropy spectrum .
in order to test this new model , we use the most recent observational data currently available . in this section , we describe how we use these data .
we use the 557 sne ia union2 dataset @xcite .
following @xcite , one can obtain the corresponding constraints by fitting the distance modulus @xmath39 as @xmath40+\mu_{0}.\ ] ] where @xmath41 , and @xmath42 is the hubble constant @xmath19 in units of 100 @xmath43 , in flat universe , the hubble - free luminosity distance @xmath44 is @xmath45 where @xmath46 . for the sn ia dataset , the best fit values of the parameters can be determined by a likelihood analysis , based on the calculation of ^2_sn=_i^557 the bao data come from sdss dr7 @xcite .
the datapoints we use are = 0.13900.0037 and = 1.7360.065 where the spherical average gives us the following effective distance measure @xcite , d_v(z)=^1/3 and @xmath47 is the comoving sound horizon at the baryon drag epoch . also , @xmath48 can be obtained by using a fitting formula @xcite : @xmath49,\end{aligned}\ ] ] with @xmath50 , \\ & & b_2=0.238(\omega_mh^2)^{0.223}.\end{aligned}\ ] ]
the function @xmath51 is the comoving sound horizon size @xmath52 where @xmath53 for @xmath54 .
so the @xmath55 for the bao data is given by @xmath56 the cmb shift parameter @xmath57 is provided by @xcite @xmath58,\ ] ] where @xmath59 is @xmath60 for @xmath61 , @xmath62 for @xmath63 , and @xmath64 for @xmath65 , respectively .
here , the redshift @xmath66 ( the decoupling epoch of photons ) is obtained by using the fitting function @xcite @xmath67\left[1+g_1(\omega_m h^2)^{g_2}\right],\nonumber\ ] ] where @xmath68 in addition , the acoustic scale is related to the distance ratio and is expressed as @xmath69.\end{aligned}\ ] ] following ref .
@xcite , the @xmath55 for the cmb data is @xmath70 where @xmath71 is a vector and @xmath72 is the inverse covariance matrix .
the seven - year wmap observations @xcite give the maximum likelihood values : @xmath73 , @xmath74 and @xmath75 . in ref .
@xcite , the inverse covariance matrix is also given as follows @xmath76 the hubble parameter can be writen as the following form : @xmath77 so , through measuring @xmath78 , we can obtain @xmath79 . in @xcite and @xcite ,
the author discuss the possibility of using absolute ages of galaxies to calculate the value of @xmath78 . in @xcite ,
the galaxy spectral data come from the gemini deep deep survey @xcite and archival data @xcite .
detailed calculations of @xmath78 can be found in @xcite , and we do not discuss them here .
currently , we have a set of 12 values of the hubble parameter versus redshift in total(see table 2 of @xcite ) .
the data calculated by this way are less sensitive to systematic errors which is a great advantage@xcite .
we can use these data to constraint different kinds of dark energy models and modified gravity models by minimizing the quantity @xmath80 this test has already been used to constrain several cosmological models@xcite .
and @xmath81 cl in the @xmath82 and @xmath83 planes for the model considered in this manuscript .
the result is consistent with the lcdm model in the @xmath5 cl .
, title="fig : " ] and @xmath81 cl in the @xmath82 and @xmath83 planes for the model considered in this manuscript .
the result is consistent with the lcdm model in the @xmath5 cl .
, title="fig : " ] in our analysis , we perform a global fitting to determine the cosmological parameters using the mcmc method . in our calculations
, we have taken the total likelihood function @xmath84 to be the products of the separate likelihoods of sne ia , bao , cmb and ohd .
then we get @xmath55 as @xmath85 the results on the best fit values of the cosmological parameters with @xmath5 confidence - level regions are : @xmath6 , @xmath7 and @xmath8 .
the nuisance parameters @xmath19 used in the analysis is actually not model parameters with significant meanings , so we do not list it . in figures [ fig : flat ] , we also show the parametric spaces @xmath83 and @xmath82 that arise from the joint analysis described above .
we note that the result is consistent with the lcdm(@xmath35 and @xmath36 ) model in the @xmath5 cl . to acquire more information on the property of dark energy , in fig .
[ fig : rev ] we plot the evolution of the eos @xmath12 along with @xmath2 .
the following is some discussion of these findings : 1 .
the best - fit results are : @xmath86 , @xmath87 and @xmath88 .
note that here the results are maximum likelihood values .
the value of @xmath12 is @xmath89 at present and @xmath12 equals to @xmath90 when @xmath14 .
we find that the best - fit dark energy model is a quintom model @xcite , whose @xmath12 crosses the cosmological constant boundary @xmath91 during the evolution . and the redshift is @xmath92 when @xmath12 crosses the cosmological constant boundary @xmath91 .
2 . with the current observational data ,
the variance of @xmath3 and @xmath4 we get are still large ; the @xmath5 constraints on @xmath3 and @xmath4 are @xmath7 and @xmath8 .
this result implies that though the dynamical dark energy models are mildly favored , the current data can not distinguish different dark energy models decisively . with the fitting results we obtained , we can reconstruct the evolution of the eos of dark energy , @xmath12 which is shown in fig . [ fig : rev ] . from the figure
, we can directly see that although the quintom model is more favored , lcdm , however , still can not be excluded .
3 . from fig .
[ fig : rev ] , one can see that the allowed value of @xmath12 is in the band @xmath93 , which is relatively narrow .
the best fit value and @xmath5 confidence - level regions of @xmath17 is @xmath94 , which is also consistent with the constraint on @xmath17 in the lcdm model and the cpl model .
along with z for the model considered in this manuscript .
the result is consistent with the cosmological constant in the @xmath5 cl . ]
in this paper , we develop a new parameterization which is the modification of efstathiou parameterization . in this new parameterization , there are three free parameters : @xmath3 , @xmath4 and @xmath17
. then we carry out the global fitting on these model using the current data : sne ia , bao , cmb and ohd . from the analysis ,
the best fit values of the cosmological parameters with @xmath5 confidence - level regions are : @xmath6 , @xmath7 and @xmath8 . from the analysis
, we can directly see that the quintom model is more favored , but this result is also consistent with the lcdm model in the @xmath5 cl .
thank xiaodong li for helping fl write some figure of this manuscript .
r. kerner , gen .
* 14 * ( 1982 ) 453 .
j. d. barrow and a. c. ottewill , j. phys .
a : math . gen . * 16 * ( 1983 ) 2757 .
j. d. barrow and s. cotsakis , phys . lett .
b * 214 * ( 1988 ) 515 .
b. li and j. d. barrow , phys .
d * 75 * ( 2007 ) 084010 .
r. r. caldwell , g. dave & p. j. steinhardt , phys .
. lett . * 80 * , 1582 ( 1998 ) b. feng , x. wang , and x. zhang , phys . lett .
b * 607 * , 35 ( 2005 ) y. wang and m. tegmark , phys .
lett . * 92 * , 241302 ( 2004 ) ; e. m. barboza and j. s. alcaniz , phys .
b * 666 * , 415 ( 2008 ) .
y. wang and p.m. garnavich , astrophys .
j. * 552 * , 445 ( 2001 ) ; c. r. watson and r. j. scherrer , phys
d * 68 * , 123524 ( 2003 ) ; p.s .
corasaniti et al . , phys .
d * 70 * , 083006 ( 2004 ) ; h.k .
jassal , j.s .
bagla and t. padmanabhan , mon . not .
* 356 * , l11 ( 2005 ) ; v. b. johri and p. k. rath , phys .
d * 74 * , 123516 ( 2006 ) .
e. m. barboza jr , j. s. alcaniz , zong - hong zhu , r. silva , phy .
rev . d. * 80 * , 043521 ( 2009 ) qing - jun zhang , yue - liang wu , jcap * 1008 * 038,(2010 ) .
qing - jun zhang , yue - liang wu , [ arxiv:1103.1953v1 ] r.a . daly and s.g .
djorgovski , astrophys .
j. * 597 * , 9 ( 2003 ) ; a. shafieloo , u. alam , v. sahni , and a.a .
starobinsky , mon . not .
astron . soc . *
366 * , 1081 ( 2006 ) .
t. holsclaw _
et al . _ ,
phys.rev.lett.*105 * , 241302,(2010 ) t. holsclaw _
et al . _ , [ arxiv:1104.2041v1 ] .
j. z. ma and x. zhang , phys . lett .
b * 699 * , 233 ( 2011 ) [ arxiv:1102.2671 [ astro-ph.co ] ] .
hong li , xin zhang , [ arxiv:1106.5658v2 ] a. friaca , j. alcaniz and j. a. s. lima , mon . not .
soc . * 362 * , 1295 ( 2005 ) .
[ arxiv : astro - ph/0504031 ] . w. hu , i. sawicki , phys .
d * 76*:104043 ( 2007 ) , arxiv:0708.1190 .
w. hu , phys . rev .
d * 77*:103524 ( 2008 ) .
w. fang et al . , phys .
d * 78*:103509 ( 2008 ) .
w. fang , w. hu , a. lewis , phys .
d * 78*:087303(2008).arxiv:0808.3125 .
e. garcia - berro , e. gaztanaga , j. isern , o. benvenuto and l. althaus , astro - ph/9907440 ; a. riazuelo and j. uzan , phys .
d * 66 * 023525 ( 2002 ) ; v. acquaviva and l. verde , jcap * 0712 * 001 ( 2007 ) .
r.gannouji and d. polarski , jcap * 0805 * 018 ( 2008 )
. w. j. percival _ et al .
_ , arxiv:0907.1660 [ astro-ph.co ] . d. j. eisenstein _ et al .
_ [ sdss collaboration ] , _ detection of the baryon acoustic peak in the large - scale correlation function of sdss luminous red galaxies , _ astrophys . j. * 633 * , 560 ( 2005 ) [ arxiv : astro - ph/0501171 ] .
d. j. eisenstein and w. hu , astrophys .
j. * 496 * 605 ( 1998 ) .
r. g. abraham _ et al .
_ , astron . j. * 127 * ( 2004 ) 2455 [ arxiv : astro - ph/0402436 ] .
j. dunlop , j. peacock , h. spinrad , a. dey , r. jimenez , d. stern and r. windhorst , nature * 381 * ( 1996 ) 581 ; h. spinrad , a. dey , d. stern , j. dunlop , j. peacock , r. jimenez and r. windhorst , astrophys .
j. * 484 * ( 1997 ) 581 [ arxiv : astro - ph/9702233 ] ; t. treu , m. stiavelli , s. casertano , p. moller , and g. bertin , mon . not .
* 308 * ( 1999 ) 1037 [ arxiv : astro - ph/9904327 ] ; t. treu , m. stiavelli , p. moller , s. casertano and g. bertin , mon . not .
astron . soc .
* 326 * ( 2001 ) 221 [ arxiv : astro - ph/0104177 ] ; t. treu , m. stiavelli , s. casertano , p. moller , and g. bertin , astrophys .
j. lett .
* 564 * ( 2002 ) l13 [ arxiv : astro - ph/0111504 ] ; l. a. nolan , j. s. dunlop , r. jimnez and a. f. heavens , mon . not .
soc . * 341 * , 464 ( 2003 ) [ arxiv : astro - ph/0103450 ] .
|
a new parameterization for the dark energy equation of state(eos ) is proposed and some of its cosmological consequences are also investigated .
this new parameterization is the modification of efstathiou dark energy eos parameterization .
@xmath0 is a well behaved function for @xmath1 and has same behavior in @xmath2 at low redshifts with efstathiou parameterization . in this parameterization
there are two free parameter @xmath3 and @xmath4 .
we discuss the constraints on this model s parameters from current observational data .
the best fit values of the cosmological parameters with @xmath5 confidence - level regions are : @xmath6 , @xmath7 and @xmath8 .
|
secant varieties have received growing attention in recent times , largely because of the fact that they provide a geometric model relevant to a wide variety of applications .
the purpose of this note is to provide a case study in finding equations of secant varieties . for an introduction to secant varieties and their applications
we invite the reader to consult @xcite and the vast collection of references therein .
the 5 factor binary secant variety is particularly interesting in light of recent results of bocci and chiantini @xcite , that @xmath3 tensors are not identifiable in rank 5 , but the generic tensor of that format has exactly 2 decompositions . for @xmath4 factors ,
the binary segre product is known to be @xmath5-identifiable for most of the possible values of @xmath5 below the generic rank , see @xcite and @xcite .
fix a field @xmath6 of characteristic @xmath7 . for @xmath8 ,
let @xmath9 be a @xmath0-dimensional vector space over @xmath6 .
let @xmath10 .
let @xmath11 be the @xmath12th secant variety of @xmath13 inside of @xmath14 ( by the segre embedding ) .
the goal of this note is to prove the following statement .
[ thm : ci ] the affine cone of @xmath11 is a complete intersection of two equations : one of degree @xmath1 , and one of degree @xmath2 . we speculate that the homogeneous coordinate ring of any secant variety of any segre product of projective spaces is cohen macaulay .
theorem [ thm : ci ] confirms this for @xmath11 . using flattening and inheritance ( * ? ? ?
* ch . 7 ) , we get the following corollary : [ cor : eqns ] suppose @xmath15 .
let @xmath16 be vector spaces , let @xmath17 and @xmath18 denote minimal generators of @xmath19 and let @xmath20 and @xmath21 , respectively , denote the natural liftings of @xmath17 and @xmath18 to @xmath22 $ ] .
let @xmath11 be the fifth secant variety of @xmath23 then the linear span of the @xmath24-orbits of @xmath20 and @xmath21 are equations that vanish on @xmath11 .
a _ flattening _ of a tensor @xmath25 is a matrix constructed by viewing @xmath26 as a linear mapping from the dual of one subset of the 5 vector spaces to the complementary subset .
the basic fact is that if @xmath26 has rank @xmath27 , then a flattening @xmath28 has rank @xmath29 .
so when non - trivial , the @xmath30 minors of flattenings provide equations of secant varieties . in our case the only possible sizes of flattenings are ( up to transpose ) @xmath31 , or @xmath32 , with maximum ranks @xmath0 and @xmath33 respectively , so they do not provide non - trivial equations for tensors of rank @xmath12 .
a next source for equations of secant varieties are exterior ( or koszul ) flattenings ( see , for instance , @xcite ) or , more generally , young flattenings @xcite . for basic background , we invite the reader to consult @xcite . the basic idea of young flattenings is to consider cases when the tensor product of a schur module @xmath34 with @xmath35 contains another schur module @xmath36 , i.e. , we have a linear map @xmath37 depending linearly on @xmath38 .
subadditivity of matrix rank implies that if @xmath39 has rank @xmath40 when @xmath26 has rank @xmath41 then @xmath39 has rank at most @xmath42 when @xmath26 has rank @xmath27 .
the art in this approach is to find good pairs of multi - partitions so that the dimensions of @xmath43 and @xmath36 are large with respect to @xmath40 , so that the young flattening has a chance to detect high rank tensors . in principle
it is possible , but tedious , to list all possible young flattenings and check which have a chance to provide meaningful equations .
because @xmath44 are all @xmath0-dimensional , there are not too many choices for @xmath45 and @xmath46 , however our initial tries at finding young flattenings that give non - trivial equations for the 5th secant variety were unsuccessful , and it seems that young flattenings do not provide equations for this secant variety . after looking for young flattenings unsuccessfully , we attempted a systematic search for equations via interpolation informed by representation theory .
our approach relied on computer calculations , which we explain in [ sec : eqn ] .
the search for the equation of degree @xmath1 is rigorous , but the search for the degree @xmath2 equation is only correct up to high probability since we only show that it vanishes on sufficiently many pseudo - random points .
our search for equations was guided by our guess that this variety , having low codimension , would be defined by just a few equations , and because of the large symmetry group , that these equations would be semi - invariants .
our guesses are validated in [ sec : proof ] , where we use these equations and one additional computer calculation on the degree of @xmath11 ( which is also only valid up to high probability ) to deduce theorem [ thm : ci ] . in [ sec : complement ] we provide a more detailed version of corollary [ cor : eqns ] from a @xmath47-module ( in the sense of @xcite ) point of view .
in particular , the equations that we find provide modules of equations for all other @xmath12th secant varieties of segre products , both when the dimensions of the factors increase ( by `` inheritance '' ) and when the number of factors increase ( by flattening ) . as far as we know these equations do not come from any known construction ( such as young flattenings ) , so they provide interesting new classes of equations for secant varieties .
we thank bernd sturmfels for suggesting this problem , j.m .
landsberg for helpful discussions , and jon hauenstein for providing bertini help .
the software bertini @xcite , macaulay2 @xcite and maple were helpful for this work .
both authors acknowledge the hospitality of the simons institute for the theory of computing in berkeley where this work was carried out .
ss was supported by a miller research fellowship .
the variety @xmath11 that we are studying has low codimension so we expect its ideal to be cut out by few equations .
in addition , the defining ideal of @xmath11 has a large symmetry group .
if one polynomial is in the ideal of @xmath11 , then so is the entire vector space of polynomials in the span of its orbit .
so we expect @xmath11 to be cut out by a small number of semi - invariant polynomials . in this section
we describe how we use all available symmetry to cut down our search for equations .
choose a basis @xmath48 for @xmath9 so that we can identify the coordinates of @xmath14 with @xmath49 where @xmath50 .
let @xmath51 denote the polynomial ring @xmath52 $ ] and let @xmath53 be the ideal of equations vanishing on @xmath11 . since @xmath51 is graded , @xmath54 , and we can compute @xmath55 for each @xmath56 .
the most naive approach to determining @xmath57 is to evaluate a basis of @xmath58 on @xmath59 points of @xmath11 using the parametrization of @xmath11 , store the results in a matrix @xmath60 and compute the kernel @xmath61 . in practice
we use pseudo - random points on @xmath11 with rational coefficients . in exact arithmetic
, non - vanishing is a certainty , but vanishing might yield a false - positive .
so this test gives an upper bound for @xmath62 and a probabilistic lower bound .
the confidence in the lower bound may be increased by evaluating on more points of @xmath11 .
alternatively , one can work over a function field over @xmath6 of large enough transcendence degree ( i.e. , use parametrized points ) where vanishing yields a genuine equation .
the downside is that such computations are more expensive .
this approach only works for small values of @xmath56 as the dimension of @xmath58 grows quickly .
in particular , one can use it for @xmath63 ( the lowest possible degree in which the equations of @xmath64 can occur , a basic fact from the theory of prolongation ( * ? ? ?
* corollary 3.4 ) ) to discover the equation @xmath65 in [ app : f6 ] , but it will not work for @xmath66 ( the largest degree we tested ) .
however , @xmath67 has an action of the group @xmath68 and we suspected that @xmath11 is defined by invariants of @xmath68 .
this gives a much smaller space of functions to search .
set @xmath69 and @xmath70 . for each even degree
@xmath71 the space of @xmath68-invariants in @xmath72 has a basis consisting of quintuples of young tableaux each of shape @xmath73 ( there are no invariants in odd degree ) .
a quintuple of tableau can be interpreted as a function on @xmath35 by applying the young symmetrizer associated to the quintuple of tableaux .
since @xmath58 is a quotient of @xmath74 , it is spanned by linear combinations of quintuples of young tableaux which now satisfy certain linear dependencies .
an explicit basis of @xmath75 may be found without explicit knowledge of all of the relations as follows .
we can verify that a given set of quintuples of tableaux are linearly independent by evaluating them on dimension - many pseudo - random points of @xmath35 .
if the matrix constructed in this way has full rank , then we have a basis of that space of invariants . if not , we continue selecting random quintuples until a basis is found .
finally , the space @xmath76 has an additional action of @xmath77 . assuming that there is a single minimal generator of @xmath78 in a given @xmath76
, it must be a semi - invariant of @xmath77 , so either an invariant or skew - invariant .
let @xmath79 ( respectively @xmath80 ) denote the subspace of @xmath81 of @xmath82-invariants ( respectively skew - invariants ) .
in figure [ ud ] we list the dimensions of these spaces of invariants for degrees up to @xmath2 .
the results follow from standard character theory calculations whose explanation we will omit .
[ ud ] [ cols="<,<,<,<",options="header " , ] we focus our search for equations of @xmath11 in the space of semi - invariants for @xmath83 .
we describe this procedure in the case @xmath66 in [ app : f16 ] .
the following proposition is a summary of what we found : [ prop : eqns ] there are minimal equations @xmath84 vanishing on @xmath11 of degrees @xmath1 and @xmath2 . both are invariant under @xmath68 .
furthermore , @xmath65 is a skew - invariant under @xmath77 while @xmath18 is a @xmath77-invariant .
to clarify : @xmath65 was constructed explicitly and we verified symbolically that it vanishes on @xmath11 and that it is a skew - invariant .
the polynomial @xmath18 was also constructed explicitly and verified to be a @xmath77-invariant , but we only verified that it vanishes on a large collection of pseudo - random points on @xmath11 , and hence it belongs to the ideal of @xmath11 with high probability . given a monomial in the @xmath49 ( the coordinates on @xmath14 ) , define its skew - symmetrization to be @xmath85 where @xmath86 is the coefficient of @xmath49 in the sum .
the polynomial @xmath65 has @xmath87 monomials and is the sum of the skew - symmetrizations of the following @xmath88 monomials : @xmath89 there is an alternative description in terms of young symmetrizers , following the same construction outlined in @xcite .
the young symmetrizer algorithm takes as input a set of fillings of five young diagrams , performs a series of skew - symmetrizations and symmetrizations , and produces as output a polynomial in the associated schur module .
one can search over all possible shapes and fillings of tableaux for a fixed number of boxes and evaluate young symmetrizers to find modules of equations in @xmath90 .
we describe this method more fully in the next section . in degree
@xmath1 the situation is particularly nice .
it turns out that there are 5 standard tableaux of shape @xmath91 and content @xmath92 and the following schur module , which uses one of each of the 5 standard fillings , realizes the non - trivial copy of @xmath93 inside of @xmath94 : @xmath95 from this description of the invariant , one can use the classical symbolic method to verify that a general point of the fifth secant variety must be a zero of this invariant ( see @xcite for complete descriptions of this type of argument , or also @xcite ) .
this gives an unconditional proof that @xmath17 is in @xmath96 .
moreover , in
[ sec : complement ] we explain how this description of @xmath17 also provides a generalization to 5th secant varieties of any larger number of segre products of projective spaces of any dimensions .
we follow the approach outlined in
[ sec : eqn - gen ] .
our search for new equations in degrees @xmath97 , @xmath98 , @xmath99 , and @xmath100 did not yield any new equations , so we only describe our process in the degree @xmath2 case .
we used maple , and the code for these computations may be found in the ancillary files accompanying the arxiv version of this paper . using the approach ,
we conclude that there are no @xmath82 skew - invariants in @xmath101 .
we found that @xmath102 and equals @xmath103 with high probability .
note that for @xmath104 , we have @xmath105 if and only if @xmath106 since @xmath65 is a skew - invariant under @xmath77 .
since @xmath107 is @xmath0-dimensional , we know that @xmath108 is a @xmath0-dimensional subspace of @xmath109 . since @xmath110 ( with high probability ) , we find that @xmath111 has one additional minimal generator in degree @xmath2 .
let us describe in more detail the case of finding @xmath82-invariants , the skew - invariant case is similar .
let @xmath112 denote quintuples of young tableaux all of shape @xmath113 .
the sum @xmath114 is in @xmath109 .
the young symmetrizer algorithm ( see @xcite ) can be used to evaluate @xmath115 on a point of @xmath35 to test if it is non - zero .
each evaluation took between 500 and 23000 seconds and up to approximately 10 gb of ram on our servers .
we continued this randomized search for non - zero elements of @xmath109 until @xmath116 linearly independent invariants were found ( the list of quintuples of fillings , data points , and code can be found in the ancillary files ) .
to verify independence , we chose @xmath116 pseudo - random points @xmath117 and verified that the matrix @xmath118 has full rank .
while each evaluation @xmath119 is expensive , they are all independent computations .
the distributed computation took approximately 2 - 3 days of computational time to verify the independence of @xmath115 .
we then took @xmath120 pseudo - random points @xmath121 and computed the matrix @xmath122 ( only @xmath116 points are strictly necessary , but we included @xmath1 more to increase the probability that our result is correct ) .
this computation took an additional 2 - 3 days to complete .
finally we found that the matrix @xmath123 has rank @xmath124 , so there is a @xmath103-dimensional space of invariants vanishing on @xmath125 and vanishing on all of @xmath11 with high probability .
an expression of a basis of the kernel of the transpose of @xmath122 gives , in turn , a basis of @xmath126 via linear combinations of quintuples of symmetrized sums of tableaux .
modding out by the space of invariants generated by @xmath17 , ( in principle ) one finds @xmath18 .
[ lem : codim2 ] @xmath11 has codimension @xmath0 .
this follows from ( * ? ? ?
* theorem 4.1 ) .
[ lem : bertini ] @xmath127 .
computing the degree of an algebraic variety is a basic function of the software bertini @xcite , see the ancillary files and @xcite .
we thank jon hauenstein for his help with this computation .
the idea is to intersect @xmath11 with a randomly chosen @xmath128 and find @xmath129 points .
one of the basic methods of numerical algebraic geometry ( and bertini , in particular ) is numerical homotopy continuation , which relies on high precision numerical path tracking to follow the paths traced out by a homotopy from a system of polynomial equations with known roots to the desired system . for more details on these types of computations applied to tensor decomposition ,
see @xcite .
the points found are represented by floating point numbers , so only satisfy the equations approximately .
though these methods have been rigorously tested through countless examples , are open source and repeatable , there is still a chance that the computations yield a false positive result , so we only claim that the result holds with high probability , so the proof of the lemma may be read as evidence for the statement . using the notation from proposition [ prop : eqns ]
, let @xmath130 , which is a complete intersection .
we have @xmath131 and @xmath132 . since @xmath11 is irreducible of codimension @xmath0 ( lemma [ lem : codim2 ] ) , and @xmath133 is equidimensional , @xmath133 is also irreducible ( otherwise the degree inequality would be violated ) .
so @xmath11 is the reduced subscheme of @xmath133 .
also , this implies that they have the same degree , so @xmath133 is generically reduced . since @xmath133 is cohen macaulay
, generically reduced is equivalent to reduced .
hence @xmath134 is a complete intersection .
as mentioned in the introduction , the equations @xmath17 and @xmath18 provide modules of equations for all other 5th secant varieties of segre products , both when the dimensions of the factors increase ( by `` inheritance '' ) and when the number of factors increase ( by flattening ) .
inheritance for secant varieties was introduced in @xcite , and was reinvestigated many times since .
inheritance can be encoded in the formality of @xmath47-modules @xcite ; we hope that our description here provides an entry point to this formalism for the unfamiliar reader .
the assignment @xmath135 is a multivariate polynomial functor @xmath51 on @xmath12-tuples of vector spaces , as is the assignment of the coordinate ring of the @xmath12th secant variety of the corresponding segre product .
similarly , there is a functor @xmath136 defined by @xmath137 ( the tor groups are @xmath138-graded , and the subscript denotes the @xmath139th homogeneous piece ) .
when @xmath140 , this is the space of minimal generators in degree @xmath139 of the ideal of @xmath141 . as a representation of @xmath142 , a @xmath143-invariant of degree @xmath144 is the schur functor @xmath145 .
so proposition [ prop : eqns ] can be interpreted as saying that @xmath146 appears with multiplicity @xmath41 in @xmath147 ( coming from @xmath65 ) and @xmath148 appears with multiplicity @xmath41 in @xmath149 ( coming from @xmath18 ) .
the koszul relation amongst @xmath65 and @xmath18 also shows that @xmath150 appears with multiplicity @xmath41 in @xmath151 .
furthermore , if any other product of schur functors @xmath152 appears in any @xmath136 then @xmath153 for some @xmath5 since it vanishes when we evaluate on @xmath154 . there is another interpretation of our results using @xmath47-modules ( see @xcite ) : the @xmath12th secant variety of the segre product of projective spaces is a @xmath47-variety , and hence the assignment of a tuple of vector spaces to the space of degree @xmath56 equations vanishing on the @xmath12th secant variety is a finitely generated @xmath47-module ( which implies finitely presented using ( * ? ? ?
* theorem 9.1.3 ) ) . for @xmath63
, we have shown that @xmath146 ( @xmath12 copies ) are minimal generators of this @xmath47-module , and similarly for @xmath66 and @xmath148 .
dustin cartwright , daniel erman , luke oeding , secant varieties of @xmath155 embedded by @xmath156 , _ j. lond . math .
( 2 ) _ * 85 * ( 2012 ) , no . 1 , 121141 , http://arxiv.org/abs/1009.1199v1[arxiv:1009.1199v1 ] .
m. v. catalisano , a. geramita , a. gimigliano , secant varieties of @xmath157 ( @xmath158-times ) are not defective for @xmath15 , _ j. algebraic geom . _ * 20 * ( 2011 ) , no .
2 , 295327 , http://arxiv.org/abs/0809.1701v1[arxiv:0809.1701v1 ] .
luca chiantini , giorgio ottaviani , nick vannieuwenhoven , an algorithm for generic and low - rank specific identifiability of complex tensors , * 35 * ( 2014 ) , no . 4 , 12651287 , http://arxiv.org/abs/1403.4157v1[arxiv:1403.4157v1 ] .
jonathan d. hauenstein , luke oeding , giorgio ottaviani , andrew j. sommese , homotopy techniques for tensor decomposition and perfect identifiability , http://arxiv.org/abs/1501.00090v1[arxiv:1501.00090v1 ] .
j. m. landsberg , giorgio ottaviani , equations for secant varieties of veronese and other varieties , _ ann . mat .
pura appl .
( 4 ) _ * 192 * ( 2013 ) , no .
4 , 569606 , http://arxiv.org/abs/1111.4567v1[arxiv:1111.4567v1 ] .
|
we describe a computational proof that the fifth secant variety of the segre product of five copies of the projective line is a codimension @xmath0 complete intersection of equations of degree @xmath1 and @xmath2 .
our computations rely on pseudo - randomness , and numerical accuracy , so parts of our proof are only valid `` with high probability '' .
|
the cosmological constant problem is by far the most severe fine - tuning problem in physics . despite many interesting proposals ( see for a review ) the problem still lacks a compelling solution .
recently , there has been an interesting idea in the context of the ` brane universe ' scenario that may solve part of the cosmological constant problem , namely the sensitivity of the effective low - energy cosmological constant to standard - model radiative corrections @xcite . in this proposal the standard - model fields
are assumed to be localized on a 3-brane in 5 spacetime dimensions .
it was argued in that special bulk - brane interactions can be chosen so that solutions with unbroken 4-dimensional poincar invariance exist independently of the standard model parameters and the 3-brane tension .
the solutions found in have a naked spacetime singularity , whose resolution in a more fundamental theory of gravity was argued to play an important role in this scenario . on the other hand ,
the description of the cancellation mechanism of the cosmological constant from a low - energy field theory perspective is not evident . also , it is not clear whether these ideas can be extended to eliminate the fine - tuning of bulk interactions @xcite . in this paper
we describe a mechanism for relaxing the cosmological constant that incorporates some of the features of , and that addresses some of these open questions .
we assume that the standard - model fields are localized on a 3-brane in 6 spacetime dimensions .
we exhibit bulk solutions with the property that no locally - defined quantity away from the 3-brane is sensitive to the value of the 3-brane tension .
this arises because the only effect of the 3-brane tension on the geometry is to induce a conical singularity in the transverse space with a deficit angle proportional to the brane tension .
these solutions have flat induced metric on the 3-brane independent of the couplings on the 3-brane , and are therefore a natural context for the ideas of . in order to obtain 4-dimensional gravity at long distances , the extra dimensions must be compactified .
we first consider several mechanisms for non - singular compactifications that can be reliably analyzed in the context of 6-dimensional gravity .
we find that none of these mechanisms preserves the property that the 4-dimensional cosmological constant is independent of the couplings on the 3-brane .
we give general reasons for this .
the solutions that we find generally have naked spacetime singularities at finite distance from the 3-brane .
these allow new compactification mechanisms that are sensitive to physics at the 6-dimensional planck scale .
we speculate that these may allow compactifications with the desired properties .
we argue that such a compactification can not be parameterized purely by higher - dimension operators , suggesting that new planck - scale degrees of freedom play an important role .
we then speculate on the behavior of gravity at long distances , assuming that a compactification mechanism with the desired properties can be found .
if the compactification scale is @xmath1 , we argue that standard model loop contributions to the effective 4-dimensional cosmological constant are cut off at distances shorter than @xmath1 .
the standard - model contribution to the cosmological constant is then of order @xmath2 , which can be small if @xmath1 is large .
this model therefore realizes the ideas of .
recent cosmological observations suggest that vacuum energy may be a significant fraction of the present density of the universe , as suggested by recent observations of type ia supernovae @xcite .
the standard - model contribution to the cosmological constant is of this order for @xmath3 mm .
as is now well - known , having 2 extra dimensions of this size do not grossly conflict with observation or cosmology @xcite .
more refined estimates of cosmological bounds give limits as strong as @xmath4 @xcite .
we will not attempt to resolve this discrepancy in the present paper , but we note that the solutions we find have a nontrivial ` warp factor ' in the bulk that can weaken the bounds .
it is also interesting that extra dimensions of this size can be probed in upcoming short - distance gravity experiments and in high - energy collider experiments .
it is important to remember that the scenario we are discussing is not a complete solution to the cosmological constant problem .
the reason is that interactions involving bulk fields and their couplings to other branes must be adjusted to special values to obtain a solution with unbroken 4-dimensional poincar invariance .
the reason that the mechanism considered here is progress is that this tuning does not involve any coupling involving the 3-brane .
this means that we can hope that the required parameter relations be made natural by unbroken symmetries in the bulk , for example supersymmetry .
it is interesting that these ideas provide a completely independent motivation for considering 2 large extra dimensions in the millimeter range .
the original motivation came from the gauge hierarchy problem @xcite .
large extra dimensions can solve the gauge hierarchy problem , and for 2 extra dimensions the size of the extra dimensions must be of order @xmath5 .
the fact that logarithmic potentials are natural in 2 extra dimensions gives a natural mechanism for stabilizing large extra dimensions @xcite .
finally , precisely in 2 extra dimensions there is a mechanism that may explain the apparent unification of standard - model couplings near the planck scale @xcite .
to describe the mechanism described above in more detail , consider a 6-dimensional metric of the form ds^2 = ^2(r ) _
dx^dx^+ dr^2 + ^2(r ) d^2 , where @xmath6 ( @xmath7 ) parameterize the space parallel to the brane , @xmath8 is the 4-dimensional minkowski metric , and @xmath9 parameterize the transverse space .
this is the most general metric with unbroken 4-dimensional poincar invariance and a translation symmetry ( @xmath10 ) in the transverse space .
we assume that the variable @xmath11 is periodic with period @xmath12 , so the translation symmetry in @xmath11 is interpreted as rotation symmetry . if @xmath13 as @xmath14 , then @xmath15 corresponds to a single point in the transverse space , like the origin of polar coordinates .
a 3-brane with extent along the @xmath6 directions at @xmath15 induces a conical singularity with deficit angle = 2= , where @xmath16 is the 3-brane tension and @xmath17 is the 6-dimensional planck scale .
away from the brane and away from other singularities , no local quantity is sensitive to the 3-brane tension . a simple way to see
this is to note that for any value of the brane tension we can rescale @xmath11 so that @xmath18 . in these coordinates , the deficit angle is parameterized by the range of @xmath11 , which can not be determined by measuring local quantities as long as there is no other deficit angle singularities away from @xmath15 .
the reason for the caveat can be understood by representing the solution with a deficit angle by a smooth metric of the form with a ` wedge ' cut out starting from @xmath15 .
specifically , we consider two radial geodesics in the transverse space emerging from the 3-brane with relative angle @xmath19 , and identify points on the geodesics that are equidistant from the 3-brane . away from the brane
the position of these geodesics is not measurable , since their absolute angular position is a coordinate artifact .
however , if the geodesics cross , the space is compactified with spherical topology , and at the second crossing there is another conical singularity with a deficit angle proportional to the deficit angle at the original 3-brane .
this must be interpreted as due to a second 3-brane , whose tension is proportional to the tension of the original 3-brane . if such singularities are not present , then there are solutions for any value of the 3-brane tension
. these solutions must be compactified . from the discussion above
, we see that we do not want to compactify the solution with a second 3-brane , since that necessarily involves tuning the two 3-brane tensions against each other .
one promising possibility is to end the space at a @xmath20 orbifold plane at constant @xmath21 surrounding the 3-brane at @xmath15 .
another possibility is that the space is not compactified , but that the warp factor @xmath22 degreases sufficiently rapidly at large @xmath21 that gravity is effectively localized at the 3-brane . to understand the limitations of these ideas , it is useful to compute the effective 4-dimensional cosmological constant assuming that a compactified solution has been obtained .
this can be obtained by integrating the 6-dimensional action evaluated at the classical solution over the transverse space .
this gives for de sitter space . ] _
4,eff = t_4 - dr d ^4(r ) ( r ) , where @xmath23 is the bulk ricci scalar , @xmath24 is the bulk cosmological constant , and the ellipses denote the contribution from possible additional bulk fields or other branes .
the ricci scalar evaluated at the solution can be written r = r_sing + r_bulk , where @xmath25 is proportional to @xmath26 , and @xmath27 contains the contributions from the bulk fields and other branes . precisely for the case of 2 extra dimensions ,
the contribution from @xmath25 exactly cancels the contribution from the first term on the of , and we have _
4,eff = -_max dr ^4(r ) ( r ) , where we have performed the ( trivial ) @xmath11 integral .
in order to solve the cosmological constant problem , the integral on the of must vanish .
this condition is completely independent of couplings on the brane , which is another manifestation of the independence of the brane tension discussed above .
however , the vanishing of the integral above is an additional condition that in general requires fine - tuning @xcite . in the present scenario , we can hope that the vanishing of this integral can be enforced by unbroken symmetries in the bulk .
for example , supersymmetry can be unbroken in the bulk , and branes away from the standard - model 3-brane may be bps states .
this is natural even if supersymmetry is broken badly by fields localized on the brane , since radiative corrections of brane fields do not renormalize the couplings associated with bulk modes and distant branes .
we begin by determining the metric away from the branes .
the solution is a special case of the one obtained in ( see also @xcite ) , but we will derive it from first principles for completeness .
we generalize the metric to allow constant curvature on the 3-brane .
we parameterize the 6-dimensional metric as ds^2 = ^2 _ dx^dx^+ s ( ) d^2 + f ( ) d^2 , where @xmath28 is a metric with constant curvature @xmath29 ( @xmath30 ) .
the ` warp factor ' @xmath31 is the radial coordinate in the transverse space , while @xmath32 is an angular variable .
the solution is particularly simple in these coordinates , as we will see .
( these coordinates are also employed in . )
the nonvanishing components of the einstein tensor are g _ & = g _ , + g _ & = g _ , + g _ & = g _ , where @xmath33 , the bulk cosmological constant and the 4-dimensional curvature @xmath29 cancel in the difference of the @xmath34 and @xmath35 equations , which give @xmath36 .
this immediately implies s ( ) = , where a possible overall constant can be set to one by rescaling @xmath6 and @xmath31 . substituting this into the @xmath35 or @xmath34 equations gives f + 3 f = - + .
the general solution is simply f ( ) = -k ^2 + + c ^-3 , where k = and @xmath37 is an arbitrary integration constant . with this solution for @xmath38 and @xmath39
, one finds that the @xmath40 equations are solved ; this is a general consequence of the reparameterization invariance in @xmath21 .
note that @xmath31 is dimensionless , @xmath41 , @xmath29 , and @xmath37 have mass dimension @xmath42 , and @xmath32 has mass dimension @xmath43 .
the relation to the more ` physical ' coordinates of is easy to work out from dr = , where the sign indicates whether the warp factor is increasing or decreasing with increasing @xmath21 . for @xmath44
, we obtain ( r ) & = ^2/5 & for @xmath45 , + ^2/5 & for @xmath46 , + ( r ) & = _ 0 (r ) , where .
we now discuss the physical interpretation of the solutions for @xmath44 . in the coordinates of
, the physical region corresponds to @xmath47 . for @xmath48
, we must have @xmath49 , and the only solution is anti de sitter space . for @xmath50 , we can choose @xmath51 by rescaling the coordinates , so that the physical region corresponds to @xmath52 for @xmath53 , and @xmath54 or @xmath55 for for @xmath49 .
for both signs of @xmath41 , @xmath56 has a simple zero at @xmath57 .
the geometry there can be exhibited by transforming to the @xmath9 coordinates : r - r_0 1 .
the metric near @xmath57 is then ds^2 _ dx^dx^+
dr^2 + r^2 d^2 , , where we have chosen the constant @xmath58 in to fix the coefficient of @xmath59 . if @xmath11 is periodic with period @xmath60 , there is no singularity at @xmath57 ( @xmath15 ) . for any other value of @xmath61
there is a deficit angle singularity at @xmath15 that we interpret as being due to the presence of a 3-brane at @xmath15 with tension proportional to the deficit angle . and solving the gravitational field equations .
we find that the deficit angle is the only effect that survives in the limit @xmath62 with the 3-brane tension held fixed . ]
the 3-brane equations of motion impose the condition @xmath63 .
we can see that this is satisfied from . for @xmath49 ,
the solution given by and for @xmath54 describes a 3-brane in infinite space of negative cosmological constant .
the ` warp factor ' @xmath31 increases away from the 3-brane , so gravity is not localized . for @xmath53
, @xmath31 decreases away from the 3-brane , and @xmath38 diverges as @xmath64 .
this is a true singularity , as can be seen from the curvature invariant r_mnpq r^mnpq 0 , where @xmath65 .
the proper distance to the singularity is @xmath66 , and the time measured by an observer on the brane for a light signal to go from the brane to the singularity and back is t = 2 _ 0 ^ 1 d .
this singularity is similar to the one found in .
it is an interesting conjecture ( following ) that in a more fundamental theory of gravity this singularity is smoothed out in a way that preserves the independence of the solution on the 3-brane tension .
for @xmath49 the region @xmath55 is also physical , although it is not possible to have a 3-brane in this region .
( note that the metric depends only on @xmath67 , so the sign of @xmath31 is not physical . )
this also has a naked singularity as @xmath64 ; in fact , the metric is the same near both singularities .
we now turn to compactification .
we first attempt to construct compactified solutions using mechanisms that we can control in the 6-dimensional gravity effective theory . we will find that these conventional mechanisms fail for very general reasons .
we then turn to possible compactifications that depend on the existence of the naked singularities .
we first attempt to find a compact solution with a 4-brane surrounding the 3-brane at fixed @xmath21 . in order to avoid a second deficit angle singularity , we assume that the 4-brane is at a @xmath20 orbifold boundary , which effectively ends the space .
it is easy to see that there are no solutions if the 4-brane action contains only a tension term .
the reason is that the 4-brane tension gives rise to boundary conditions ( in the metric of ) = = - . at a symmetric point
, this implies that @xmath68 on both sides of the 4-brane .
if we use this as initial data for evolution in @xmath21 , we obtain a solution with @xmath69 .
this can not evolve in a finite distance to the solution near the 3-brane , which has @xmath70 .
we now consider the possibility that there is a non - minimal stress tensor on the 4-brane of the form t _
= -t_5 _ , t _ = -t_5 , _ , where @xmath71 is the induced metric on the 4-brane .
this is the most general stress tensor compatible with the symmetries we are assuming .
the asymmetry @xmath72 allows us to find solutions with a 3-brane surrounded by a 4-brane at finite distance , as we will see below . in the limit
@xmath73 the proper separation between the 3-brane and the 4-brane becomes infinite , so the mechanism that gives rise to the asymmetry is crucial for the compactification . to cancel the standard - model contributions to the cosmological constant ,
it is important that the mechanism that gives rise to the asymmetry does not bring in dependence on the 3-brane tension via the deficit angle .
for example , casimir energy from fields localized on the 4-brane will give rise to a stress tensor of the form , but @xmath74 is proportional to the circumference of the 4-brane , which depends on the deficit angle at the 3-brane .
similarly , a scalar field localized on the 4-brane that wraps in the @xmath11 direction must come back to its original value around the 4-brane , and hence its stress tensor also depends on the deficit angle .
a more promising mechanism is to localize a 3-form gauge field on the 4-brane and assume that it has nonzero field strength f _ = e _ , e = , with all other components vanishing .
this is a constant ` electric ' field in the @xmath6 directions , and is not subject to any quantization condition .
the sign of the stress tensor from this configuration gives @xmath75 .
we will see that this is the sign that we require to obtain a solution .
the solution requires a single fine - tuning involving @xmath76 , but this fine - tuning does _ not _ involve the 3-brane tension . nonetheless , this is not a satisfactory solution . to understand the reason ,
consider adiabatically changing the 3-brane tension .
this can be thought of as a crude model for a phase transition involving matter on the 3-brane .
the metric then has the form with an adiabatically evolving deficit angle .
as the deficit angle changes in time , the size of the 4-brane changes .
this changes the value of @xmath76 , since it is the integrated number of field lines ( rather than @xmath76 ) that is conserved in the adiabatic evolution .
the 4-form flux mechanism has effectively recast the cosmological constant problem as a problem of fine - tuning initial data .
this phenomenon is actually a general consequence of the conservation of stress - energy . to see this ,
consider a stress tensor of the form in a metric with adiabatically changing deficit angle .
this metric can be parameterized by the replacement @xmath77 in ( with a fixed range for @xmath11 ) .
the equation @xmath78 then implies _
t t_5 = ( t_5 , - t_5 ) .
we see that @xmath79 must change in response to the deficit angle . in the remainder of this section
, we construct the solutions explicitly .
the main goal of this is obtain a quantitative estimate of the fine - tuning of the 4-brane tension that is required to obtain a small cosmological constant .
we find that there is no improvement relative to a generic theory with new physics at the tev scale .
the reader who is not interested in these details can skip to the next subsection .
away from the 4-brane , the solution has the form considered in the previous section . at the 4-brane ,
the metric components have discontinuous first derivatives . in terms of the metric ,
the relevant terms in the einstein tensor are those with second derivatives with respect to @xmath21 .
these are g _ & = g _ , + g _ & = g_. the 4-brane gives a nonvanishing contribution to the @xmath40 and @xmath80 components of the stress tensor proportional to @xmath81 , where @xmath82 is the radial position of the 4-brane .
this gives the discontinuity conditions = - , =
-t_5 + , where we now use units where @xmath83 .
the functions @xmath31 and @xmath84 are continuous at the 4-brane .
it is convenient to write these conditions in terms of the ` warp factor ' coordinates using = , = 1 .
the sign factor @xmath85 tells us whether the warp factor is increasing or decreasing with increasing @xmath21 ; it can be chosen independently on either side of the brane .
we choose the coordinate @xmath21 to increase monotonically as we pass through the 4-brane from the 3-brane position at @xmath15 . because of the @xmath20 orbifold projection
, we require the solution to be symmetric with respect to reflections about the 4-brane .
this means that the bulk solution is described by the same function @xmath38 on both sides of the 4-brane .
the @xmath86 discontinuity equation is then ( _ 2 - _ 1 ) = - , f_0 f(_0 ) , where @xmath87 is the warp factor at the 4-brane .
we first consider a space with negative cosmological constant .
the warp factor increases away from the 3-brane , so @xmath88 , @xmath89 . from
we see that this requires @xmath90 .
can then be written as -k + _ 0 ^ -2 + c _ 0 ^ -5 = ( ) ^2 . using = ,
the @xmath91 discontinuity equation can be written - = -t_5 + .
combined with , this can be simplified to give -2 k - 3 c _ 0 ^ -5 = ( t_5 - t_5 , ) . for @xmath44 , we can combine this with to obtain k = - ( t_5 - t_5 , ) .
this is a fine - tuning condition that is required to obtain a solution with vanishing 4-dimensional curvature . for @xmath92 , and can be solved for @xmath87 and @xmath29 . a similar fine - tuning in
is required to ensure the flatness of a brane with codimension 1 .
in fact , if we take the limit @xmath93 , @xmath73 , we obtain exactly the solution of in one dimension higher .
we now assume that @xmath50 , so that we can rescale coordinates and choose @xmath51 .
we also set @xmath44 .
the solution then has a deficit angle at @xmath57 that we interpret as a 3-brane .
the 4-brane position @xmath94 is then determined by and to be _ 0 ^ -5 = ( t_5 , - t_5 ) . the proper radial distance between the 3-brane and 4-brane is = _
1^_0 = . as discussed above , in the limit @xmath73
, we have @xmath95 and @xmath96 . in order to obtain finite @xmath1 with @xmath97
, we need @xmath98 ( remember @xmath49 ) .
these signs are compatible with the stress tensor from a 3-form gauge field , as discussed previously .
the effective 4-dimensional planck scale is given by m_4 ^ 2 = ( _ 0 ^ 3 - 1 ) _ max .
the warp factor has maximum value at the 4-brane , so the solution tends to localize gravity there .
the warp factor grows exponentially as a function of @xmath99 ( see ) , so we can solve the gauge hierarchy problem via the mechanism of for @xmath100 .
we will not pursue this possibility here . for @xmath53
, a similar analysis shows that and hold in this case as well , but now @xmath101 and @xmath102 . the fact that the 4-brane has negative tension is not a problem , since it is fixed on the orbifold plane . in these solutions , the warp factor decreases away from the 3-brane , and the 4-brane cuts of the space before the naked singularity at @xmath103 .
we now consider the fine - tuning of the cosmological constant in these solutions . if we perturb the 4-brane tensions away from their fine - tuned values , @xmath104 , @xmath105 , we obtain & = + ( t^2 ) , + _ 0 & = - + ( t^2 ) .
we estimate the 4-dimensional curvature @xmath29 for an order-1 perturbation of the 3-brane tension @xmath106 .
this results in a change of the deficit angle @xmath107 , which may be small if @xmath108 .
( this can be natural due to approximate susy on the 3-brane . ) also , @xmath109 and @xmath110 are naturally of order @xmath111 , where we allow @xmath112 .
this results in a 4-dimensional curvature ~. combining this with , , and , we obtain . since @xmath113 ,
this is no better than a 4-dimensional theory with a tev scale vacuum energy ( a theory with supersymmetry broken at the tev scale ) .
we now consider the possibility that the extra dimensions are infinite but with a ` warp factor ' @xmath22 that decreases sufficiently rapidly as @xmath114 so that gravity is approximately 4-dimensional at long distances @xcite . in this scenario , since gravity is not 4-dimensional even at arbitrarily long distances , it is obvious how it evades ` no - go ' theorems concerning the tuning of the cosmological constant @xcite . the simplest possibility would be to surround the 3-brane by a 4-brane in the extra dimensions , with the spacetime being anti de sitter on the outside of the 4-brane .
however , the anti de sitter metric has the symmetry @xmath69 , which is preserved by the brane matching condition .
therefore , anti de sitter space can not match onto the asymmetric solution that is required if we have a 3-brane in the solution .
another possibility is to have additional fields in the bulk , and look for solutions where the warp factor vanishes at infinity .
this can evade the obstruction described above , since an asymmetric solution with @xmath70 can evolve to a symmetric solution asymptotically .
the simplest possibility is to introduce scalars @xmath115 into the bulk : s_scalar = d^6 x .
the resulting einstein equations can be simplified by introducing the quantities ( r ) _ r = + , ( r ) - .
we require @xmath116 to decrease as @xmath114 to have finite volume , so @xmath117 .
@xmath118 is a measure of the asymmetry between the @xmath11 and @xmath6 directions . in terms of @xmath119 and @xmath118 ,
einstein s equations are first - order differential equations .
taking the difference of the @xmath40 and @xmath80 components of einstein s equations gives = - . since @xmath117 , we see that @xmath120 increases monotonically as @xmath114 .
this means that we can not hope to obtain a solution that approaches anti de sitter space ( which has @xmath69 and hence @xmath121 ) with a decreasing warp factor . we may still hope to find compactified solutions with a decreasing warp factor that are not anti de sitter at infinity . using to simplify the difference of the @xmath80 and @xmath122 einstein equations gives
= -^2 - _a _a .
because the is negative - definite , @xmath119 decreases ( becomes more negative ) monotonically .
it can be shown that all solutions with a decreasing warp factor have a singularity at finite @xmath21 by using the fact that the solutions are bounded by the solutions with @xmath123 . in detail , let @xmath124 and @xmath125 be solutions with some initial conditions at @xmath126 , and let @xmath127 and @xmath128 be the solutions of and with the same initial conditions , but setting @xmath123 .
we then have @xmath129 and hence @xmath130 for @xmath131 .
it is not hard to see that @xmath132 at finite @xmath21 , so @xmath125 must also have a singularity at finite @xmath21 .
a striking feature of the solutions constructed in the previous section is the presence of naked singularities . near a naked singularity the curvature is blowing up ,
so the physics of the singularity is sensitive to the details of physics above the planck scale .
it is natural to speculate ( following ) that this new physics ends the space at the singularity . in this scenario , the long - distance behavior of gravity
is directly controlled by the short - distance behavior at the singularity , and we can not rigorously address the physics of the singularity using the low - energy effective theory .
we will therefore confine ourselves to some simple observations .
first , we note that the curvature distinguishes between the @xmath6 and @xmath11 directions .
( for example , @xmath133 for @xmath7 . )
this is potentially important for compactification because the failure of the compactification mechanisms above can be traced to the non - existence of terms in the equations of motion that distinguish between the @xmath6 and @xmath11 directions independently of the 3-brane tension .
near the naked singularity , these higher - derivative effects may be important and allow more general boundary conditions than the ones we considered .
the resulting theory would be a @xmath20 orbifold with planck - scale curvature near the orbifold boundary .
we can attempt to get some insight into this scenario by adding higher - derivative terms to the action .
we can not hope to obtain consistent solutions , but we can see that the hoped - for effects do not occur if we treat the higher - derivative terms perturbatively . at each order in perturbation theory , we have a system of second - order equations that involves the lower - order solutions . at each order , the boundary conditions on the metric at the orbifold boundary
is @xmath134 , which gives rise to a symmetric solution ( @xmath135 ) . for higher - derivative terms localized on the orbifold boundary
, the same argument holds if we regulate the brane ( by a scalar domain wall ) . if we attempt to include the higher - derivative terms non - perturbatively , the initial data at the orbifold boundary involves higher derivatives of the metric . taking asymmetric initial conditions
, one can presumably find solutions of the kind we seek .
however , it is difficult to interpret such solutions physically .
the additional initial conditions can be thought of as additional planck - scale degrees of freedom .
however , theories with higher - derivative terms are generally classically unstable , corresponding to the fact that the extra degrees of freedom are generally ghosts . at best
, this line of reasoning may be viewed as a weak hint that new planck - scale degrees of freedom localized at the orbifold boundary may allow compactification .
recently , there has been an interesting proposal to make sense out of naked singularities such as this by imposing boundary conditions at the singularity @xcite .
while we regard this proposal as very interesting , we note that it appears to be difficult to give a generally covariant formulation of the boundary conditions for the metric ( see however ) .
another interesting approach is to look for solutions that ` regulate ' the singularity by hiding it behind an event horizon @xcite .
these approaches are definitely worthy of further exploration .
we conclude that there are potential difficulties with the idea of compactification near the naked singularities , but the idea can not be ruled out . in light of this , we believe that the scenario described here is worth further investigation .
we now turn to the effective field theory analysis of the scenario described above , assuming that compactification is possible . without a specific compactification mechanism
, we can not address the details of the kaluza - klein ( kk ) spectrum or the 4-dimensional effective field theory , but we can analyze some simple aspects of the scenario that depend on the behavior near the 3-brane .
we first consider the light degrees of freedom .
the kk spectrum contains the 4-dimensional graviton and a massless gauge field corresponding to the unbroken rotational invariance in the transverse space .
these can be parameterized by the metric ds^2 = ^2(r ) g_(x ) dx^dx^+ dr^2 + ^2(r ) ^2 , where @xmath22 and @xmath136 are as in our solution .
the kk gauge field couples to momentum in the @xmath11 direction , so the standard - model fields are not ` charged ' under the gauge group .
in addition , the couplings of the kk gauge field is suppressed by large volume factors and the wavefunction factor @xmath136 above .
the bounds on the couplings of such a vector are therefore much weaker than the corresponding bounds on kk gravitons , which is safe .
another important question we can address is whether higher - order terms on the brane upset the non - dependence of the 4-dimensional cosmological constant on the 3-brane tension .
for example , we can write s_brane ~d^4 x m_6 ^ 2 .
such terms can be generated by standard - model loops with external gravitational lines ; in fact , _ all _ such loop effects proportional to positive powers of @xmath17 ( the cutoff ) correspond to ultraviolet divergences , and therefore correspond to operators localized on the brane .
since all such effects are equivalent to tree - level terms , this reduces the question to a classical analysis . making sense of the equations of motion that follow from requires regulating the 3-brane , presumably taking care to preserve general covariance .
however , a very general argument shows that the independence of the 4-dimensional cosmological constant of the 3-brane tension is robust against the addition of such effects .
let us momentarily adopt the contrary hypothesis , namely that the absence ( or fine - tuning ) of the couplings in are required to obtain a solution with unbroken 4-dimensional poincar invariance .
we should then ask what the solutions are . if we violate the fine - tuning of the couplings away from the 3-brane by a small amount , it is clear that the effect in the 4-dimensional effective field theory is a nonzero cosmological constant . the vacuum solutions in the 4-dimensional field theory are then de sitter or anti de sitter space .
the 6-dimensional metric that corresponds to this solution should therefore have the symmetries of 4-dimensional de sitter or anti de sitter space ; this metric was given in , , and , where the parameter @xmath29 is the 4-dimensional curvature .
this metric admits deficit angle singularities at simple zeros of @xmath38 .
since these are the _ only _ solutions with the required symmetries , we conclude that all the effects of the 3-brane couplings can be absorbed into the deficit angle and the 4-dimensional curvature .
( this is analogous to a ` no - hair ' theorem for codimension 2 branes . )
we must still address the possibility that the 4-dimensional curvature is sensitive to the terms in .
when these terms are properly regulated , their effect on physics below the scale @xmath17 can be written as boundary conditions on the gravitational fields involving higher derivatives . if the terms in can be treated as non - singular perturbations , these new boundary conditions will still allow solutions for any value of the 4-dimensional curvature , including flat space . under our assumptions
, it is the matching condition at the _ other _ boundary of the transverse space that picks out the value of @xmath29 .
we conclude that the presence of the terms does not invalidate our picture .
this argument eliminates contributions to the effective 4-dimensional cosmological constant proportional to positive powers of @xmath17 .
however , we expect loop matching corrections from loops of standard - model fields of order _ 4,eff ~. the scale of these corrections is set by @xmath1 because the size of the dominant loops in position space is of order @xmath1 . there are no contributions involving positive powers of @xmath17 because these would have to correspond to a local counterterm in the 6-dimensional theory that gives an @xmath1-dependent contribution to @xmath137 at tree level . to the effective 4-dimensional cosmological constant , which will be much larger than for large @xmath138 .
however , the different kk states are nothing more than different eigenstates of momentum in the compact directions , and so such contributions correspond to contributions proportional to the cutoff @xmath17 from loops with high momentum .
the argument above shows that these can not occur .
the underlying reason is 6-dimensional locality , which is not manifest in a kk description .
] we can ask whether it is possible that @xmath137 could be _ smaller _ than the estimate .
we believe that this is impossible , simply because of the naturalness of the effective theory at distances larger than @xmath1 .
there are no massless scalars in this effective theory , and hence no light degrees of freedom that can adjust the cosmological constant to zero .
we can get some insight into the mechanism for the cancellation of the standard model contribution to the cosmological constant by considering the dynamics of a slowly rolling scalar field localized on the 3-brane .
the potential of the scalar acts as an effective 3-brane tension that varies with time .
if the rate of change of this tension is sufficiently slow , the bulk gravitation fields will respond adiabatically .
the solution will therefore have a deficit angle that tracks the instantaneous value of the scalar potential .
from the point of view of the 4-dimensional effective theory , this corresponds to a mixing between the scalar and the gravity kk mode that corresponds to the deficit angle .
this light mode adjusts itself to cancel the cosmological constant .
we expect this mode to have a mass of order @xmath139 , and below this scale the cancellation mechanism is no longer effective ; this leads to the estimate .
however , we do not understand the generation of a nonzero cosmological constant from a 6-dimensional perspective .
the arguments above address only the contribution of standard - model loops to the effective cosmological constant .
we emphasize again that even if one accepts the existence of a compactification with the properties described in the previous section , we do not have a complete solution to the cosmological constant problem .
in particular , we have not addressed the question of bulk gravity loops , and we have seen that fine - tuning of bulk interactions is necessary to obtain a solution with unbroken 4-dimensional poincar invariance .
the important point is that the quantities that must be fine - tuned do not involve the 3-brane couplings .
we can therefore hope that unbroken bulk symmetries such as supersymmetry can make these parameter choices natural .
we therefore believe that it is plausible that the leading contribution to the cosmological constant is of order .
we have described a natural mechanism for canceling the standard - model contribution to the cosmological constant .
it relies only on the properties of branes with 2 transverse dimensions .
the mechanism requires the compactification scale to be in the millimeter range , and suggests a nonzero cosmological constant in the range favored by cosmology .
previous work has shown that the presence of 2 large extra dimensions can also explain the gauge hierarchy problem and the unification of gauge couplings ; this confluence of ideas is nothing if not suggestive .
most importantly , these ideas are testable by terrestrial experiments and cosmological observations .
while we have not definitely established all aspects of the mechanism we proposed , we hope that some of these ideas will prove fruitful in the search for the ultimate solution of the cosmological constant problem .
we thank a. nelson and r. sundrum for important comments on an earlier version of this paper .
we also thank n. arkani - hamed , t. jacobson , r. sundrum , and j. terning for discussions .
this work was supported by the national science foundation under grant phy-98 - 02551 , and by the department of energy under grant doe / er/40762 - 204 .
|
we consider ` brane universe ' scenarios with standard - model fields localized on a 3-brane in 6 spacetime dimensions .
we show that if the spacetime is rotationally symmetric about the brane , local quantities in the bulk are insensitive to the couplings on the brane .
this potentially allows compactifications where the effective 4-dimensional cosmological constant is independent of the couplings on the 3-brane .
we consider several possible singularity - free compactification mechanisms , and find that they do not maintain this property .
we also find solutions with naked spacetime singularities , and we speculate that new short - distance physics can become important near the singularities and allow a compactification with the desired properties . the picture that emerges is that standard - model loop contributions to the effective 4-dimensional cosmological constant can be cut off at distances shorter than the compactification scale . at shorter distance scales ,
renormalization effects due to standard - model fields renormalize the 3-brane tension , which changes a deficit angle in the transverse space without affecting local quantities in the bulk . for a compactification scale of order @xmath0
, this gives a standard - model contribution to the cosmological constant in the range favored by cosmology .
|
in this supplementary material we first show that the inter - particle interaction of the @xmath35 shape , used throughout the present study , appears naturally in the context of the `` rydberg - dressed '' atoms .
so , it is possible to verify all the predicted physics within presently available experimental setups .
next , we demonstrate how to control the number of humps and fragmentation ratio in the ground state of the bosonic system with such an inter - particle interaction trapped in simple barrier - less traps , also see ref .
we also show how to distinguish fragmented systems studied here from the condensed ones , by means of the correlation functions .
finally , we provide the full movies of the time - dependent evolutions in 2d and 3d setups , discussed in the main text .
let us start from the issue of possible experimental realizations of the predicted physics .
to observe the predicted effects the first necessary condition is that the length / range of the inter - particle interaction should be comparable with the length of trapping potentials .
the second necessary condition is that the interaction should be strong enough .
we expect to observe the formation of few - hump fragmented ground states in trapped ultra - cold systems made of `` rydberg - dressed '' atoms [ 5 - 7 ] .
let us discuss how to realize the predicted multi - hump muti - fold fragmented physics in rydberg - dressed " systems .
the off - resonant optical coupling of ground state atoms to highly excited rydberg states [ 6 , 7 ] allows one to modify and control the shape and strength of effective two - body interactions .
the advantage of this technique is that a small component of the rydberg state is admixed to the ground state of atoms , providing thereby an additional degree of manipulation of the interaction strength .
the shape of the potential energy of the two many - electron atoms excited to a rydberg state depends very much on details of the electronic structure .
the most common long - range behavior , however , is of a standard van der waals ( @xmath76 ) type with possible admixture of terms of other degrees , e.g. , a pure dipole interaction ( @xmath77 ) .
these additional contributions are responsible for the tails of the interaction potentials .
the dressing " of two rydberg atoms with van der waals type of interaction results , see refs .
[ 6 , 7 ] , in the two - body effective inter - particle potential : @xmath78 let us make a change of variables @xmath79 , where @xmath33 is the frequency of the external trap . in the units of the external confining potential s length , @xmath80
, the inter - particle interaction reads @xmath81 where @xmath82 is the rescaled interaction strength and @xmath83 . in the present study ,
we have used one- , two- and three - dimensional versions of inter - particle interactions of a similar shape @xmath84 of half - width @xmath36 with @xmath37 .
this degree of the inter - particle repulsion function corresponds to an `` intermidiate '' situation between the pure dipole - dipole and van der waals interactions .
we have used it to demonstrate the generality of the observed physics , which holds for @xmath85 and @xmath86 as well .
the strength @xmath2 of the inter - particle interaction depends on experimentally tunable parameters : @xmath87 a two - photon rabi frequency between the involved atomic levels , @xmath88 detunings of lasers with respect to the atomic transitions , and the external confinement @xmath33 .
the screening " constant @xmath89 defines the critical distance , below which the interaction is constant , i.e. , originates to the blockade phenomenon [ 5 ] .
this range depends on the detuning @xmath88 and on the pure spectroscopic @xmath90 coefficient governed by the electronic structure of the excited state , i.e. , it can be manipulated by a proper choice of an atomic rydberg level .
what are the presently available / reachable experimental conditions for rydberg excitations ? in recent experiment , see ref .
[ 8 ] , with @xmath91rb bose - einstein condensates , a few tens of rydberg atoms in a quasi-1d trap have been successfully detected .
the reported blockade radii were between 5- while the radial dipole trap frequencies used were around , resulting in a radial length of order of 1- , i.e. , a ratio @xmath92 , where @xmath93 is the screening " constant of the rb atoms .
it means that the range of the interaction was larger than the the size of the trap in the transverse direction .
the strength of the interaction of the dressed rydberg atoms is defined by the @xmath94 ratio and can be tuned between 0.1- , implying that in a trap the corresponding dimensionless interaction strength @xmath82 can be @xmath95 1 - 100 .
so , formally , the lengths / ranges and interaction strength needed to observe the predicted multi - hump muti - fold fragmented states and their dynamics are already reachable with ultra - cold dressed " rydberg atoms .
let us now demonstrate that ground state properties of the trapped bosonic system with a finite - range repulsive interaction can be controlled by varying either the strength of the inter - particle repulsion @xmath2 , the tightness / strength of harmonic confinement @xmath33 , or by changing the number @xmath34 of the trapped particles . for more details see ref .
[ 10 ] . here
, we explicitly consider 1d bosons with finite - range inter - particle interaction function @xmath96 with @xmath36 .
the ground state of @xmath7 bosons trapped in a standard @xmath97 harmonic trap , i.e. , with @xmath98 , at weak repulsion @xmath99 is super - fluid , i.e. , condensed and has the one - hump density , see fig .
s1(a ) . by increasing the strength of the repulsion one
gets at @xmath100 a two - fold fragmented ground state with two - hump density , see fig .
s1(b ) . at even stronger repulsion ,
@xmath101 , the ground state is three - fold fragmented with a three - hump density profile , see fig .
alternatively , three - fold fragmented states with a three - hump density profiles can be prepared in a weaker external trap @xmath102 for @xmath7 with @xmath103 , see fig .
s1(f ) or in a harmonic trap with @xmath98 , but for @xmath104 and @xmath105 , see fig .
s1(d ) . finally , fig .
s1(e ) shows that the system of @xmath7 with @xmath103 placed in an asymmetric trap @xmath106 , results in a two - fold fragmented ground state with an asymmetric two - hump density , verifying thereby that the symmetry of the trap does not play a significant role here . , the tightness / strength of external confinement @xmath33 , or by changing the total number @xmath34 of the trapped particles .
the ground state of the system with @xmath7 interacting via a finite - range @xmath96 with @xmath36 trapped in a standard @xmath97 harmonic trap , i.e. , with @xmath98 , at weak repulsion @xmath99 is super - fluid and has the one - hump density ( a ) . at stronger interactions it becomes multi - fold fragmented with two - hump densities at @xmath100 ( b ) and three - hump at @xmath101 ( c ) .
alternatively , the three - hump densities can be obtained by preparing the system of @xmath104 with @xmath105 in the standard harmonic trap ( d ) , or the system of @xmath7 with @xmath103 in a weaker external trap with @xmath102 ( f ) .
the system of @xmath7 with @xmath103 placed in an asymmetric trap @xmath106 results in an asymmetric density of the fragmented state ( e ) .
all quantities shown are dimensionless.,width=377 ] the above demonstrated versatility of control functions implies that in experiments a desired fragmentation scenario can be equivalently obtained ( i ) by squeezing / opening the external trap ; ( ii ) by changing the number of trapped bosons ; or ( iii ) by varying the strength of the repulsion .
the time - dependent studies reported in the main text have equipped us with detailed time - dependent understanding of the above mentioned control functions .
namely , the topology of a once prepared system can not be easily and smoothly changed by a simple , sudden quench of the repulsion .
indeed , we have shown that attempts to bring an initially two - fold fragmented system to the super - fluid , one - hump regime via strong decrease of the interaction , or to a three - fold fragmented state via strong increase of the interaction have resulted in explosive dynamics .
the systems with different number of humps have different topologies and live in different configurational ( fock ) spaces , implying that to bring one system to another one we have to `` transfer '' a macroscopic number of the particles from one fragment ( hump ) to the other fragments which is , clearly , a very non - trivial many - body task .
furthermore , it is easier to realize the systems with a few humps and many particles per hump than the many - hump system with a few particles per hump . since the height of the induced barrier is proportional to the interaction strength and to the number of particles residing in the fragment , the humps with a larger number of particles are more stable with respect to exchange and loss of particles , i.e. , to the temperature .
now we would like to demonstrate that all the multi - hump systems studied here are initially fragmented and remain fragmented during the non - violent and explosive time - evolutions . in fig .
s2 we depict the one - body correlation functions @xmath107 computed for the 1d dynamical scenarios of fig .
1(a , b , c ) at time - slice of @xmath108 .
these correlation functions have typical characteristic patterns of two - fold fragmented structures , see [ 18 ] .
the bright squares in fig .
s2(a ) mean that , within each sub - cloud , the bosons are coherent ( condensed ) , while the dark off - diagonal squares imply that there are no definite phase relations ( correlation ) between the sub - fragments , pretty much as in mott - like states [ 19 ] . in fully condensed systems , in contrast , the phase between different humps is fixed , i.e. , the correlation function of the evolving fully condensed system would always be flat and bright everywhere , i.e. , @xmath109 . taken at the same time - slice @xmath110 for the 1d dynamical scenarios depicted in fig .
1(a , b , c ) of the main text .
the two - fold fragmented structures of the evolving wave - packets survive during the non - violent ( a , b ) and explosive ( c ) dynamics .
all quantities shown are dimensionless.,width=321 ]
2d : : ` 2d_mctdhb_quench_to_l_0d1.mpg ` file contains the movie exemplifying the over - a - barrier evolution of a two - hump two - fold fragmented initial state in a 2d setup induced by a sudden displacement of the trap @xmath59 with a simultaneous strong reduction of the inter - particle repulsion @xmath42 .
one slide of this movie , corresponding to a snap - shot at @xmath50 , is depicted in fig .
2d : : ` 2d_mctdhb_quench_to_l_0d7.mpg ` file contains the movie exemplifying the under - a - barrier evolution of the above scenario but with a moderate increase of the inter - particle repulsion @xmath51 .
one slide of this movie , corresponding to a snap - shot at @xmath52 , is depicted in fig .
2d : : ` 2d_mctdhb_quench_to_l_0d8.mpg ` file contains the movie exemplifying the over - a - barrier evolution of the above scenario but with a stronger increase of the inter - particle repulsion @xmath53 .
one slide of this movie , corresponding to a snap - shot at @xmath54 , is depicted in fig .
3d : : ` 3d_mctdhb_quench_to_l_0d1.mpg ` file contains the movie exemplifying the over - a - barrier evolution of a two - hump two - fold fragmented initial state in a 3d setup induced by a sudden displacement of the trap @xmath111 and strong sudden decrease of the repulsion from @xmath42 . in fig .
4 four slides of this movie , corresponding to snap - shots at @xmath71 , are depicted .
3d : : ` 3d_mctdhb_quench_to_l_0d7.mpg ` file contains the movie exemplifying the under - a - barrier evolution of the above 3d scenario but with a moderate increase of the inter - particle repulsion @xmath51 .
3d : : ` 3d_mctdhb_quench_to_l_0d8.mpg ` file contains the movie exemplifying over - a - barrier evolution of the above 3d scenario but with a stronger increase of the inter - particle repulsion @xmath53 .
|
two generically different but universal dynamical quantum many - body behaviors are discovered by probing the stability of trapped fragmented bosonic systems with strong repulsive finite / long range inter - particle interactions .
we use different time - dependent processes to destabilize the systems a sudden displacement of the trap is accompanied by a sudden quench of the strength of the inter - particle repulsion .
a rather moderate non - violent evolution of the density in the first `` topology - preserved '' scenario is contrasted with a highly - non - equilibrium dynamics characterizing an explosive changes of the density profiles in the second scenario .
the many - body physics behind is identified and interpreted in terms of self - induced time - dependent barriers governing the respective under- and over - a - barrier dynamical evolutions .
the universality of the discovered scenarios is explicitly confirmed in 1d , 2d and 3d many - body computations in ( a)symmetric traps and repulsive finite / long range inter - particle interaction potentials of different shapes .
implications are briefly discussed .
one of the most bright universal features shared by many - body systems with strong repulsive interaction is the formation of multi - hump structures in the ground states densities .
driven by the strong repulsive interaction they can be formed in the systems with short- and finite / long - range inter - particle interactions in one- ( 1d ) , two- ( 2d ) and three - dimensional ( 3d ) setups .
the famous examples in the context of ultra - cold systems are strong contact interaction and tonks - girardeu gases in 1d @xcite and in more general physical contents the formation of super - solids and crystals in 2d systems with long - range interactions @xcite . while ground state properties of these systems have been accessed at different levels of the quantum theory , the many - body studies on excited states , needed to digest dynamical behavior and stability of these systems are rather scarce . because of the intrinsic complexity and correlations of these systems , the understanding of their dynamical stability as a time - dependent process at a proper many - body level is a challenging theoretical and cumbersome computational task and is still missed . in this letter
we investigate the stability of such strongly repulsive many - boson systems with muti - hump many - body states as a time - dependent process by solving the time - dependent many - boson schrdinger equation @xmath0 in 1d , 2d and 3d setups for several dynamical scenarios involving manipulations with external trap @xmath1 and with the strength @xmath2 of inter - boson interaction potential @xmath3 .
the respective many - body hamiltonian is @xmath4 \!+\!\sum_{j < k}^n \lambda_0 w(\r_j\!-\!\r_k ) , \nonumber$ ] @xmath5 , @xmath6 .
all the results reported in this letter have been obtained for @xmath7 bosons interacting via inter - particle interaction function @xmath8 of half - width @xmath9 with @xmath10 in 1d , 2d and 3d .
the inter - particle interactions of similar shapes naturally appear in the so - called `` rydberg - dressed '' ultra - cold systems @xcite which are of current experimental interest @xcite .
the repulsive inter - particle interaction functions of other shapes with similar range and strength would result in qualitatively the same physics as reported here , see also the supplemental material @xcite for discussion .
recently , we have verified at an accurate many - body level @xcite that in bosonic systems with strong repulsion confined in simple , barrier - less traps particular patterns of the ground state density , i.e. , number of the humps and correlations , are governed by the geometrical interplay between the width / range of the inter - boson interaction potential and the available volume ( length of the trap ) .
examples of such two - hump states in 1d and 3d setups are depicted in fig .
[ fig1 ] and fig .
[ fig3 ] at t=0 . intuitively , one can associate each hump of a multi - hump state with a localized subsystem and to describe , therefore , the overall wave - function @xmath11 as a superposition of all constituting fragments . a configuration where @xmath12 bosons are residing in the first fragment @xmath13 , @xmath14 in @xmath15 , etc . , and @xmath16 in @xmath17
is called permanent @xmath18 : @xmath19 . here
we have already admitted that the shapes of the constituting sub - clouds can evolve in time . for a realistic description this idealized picture of a single configuration with a fixed number of particle residing in each fragment should be augmented by other processes describing , e.g. , the exchange and hopping of particles between the fragments . the many - body theory naturally taking into account this time - dependency of the fragments as well as all possible hopping processes within has recently been developed and called multi - configurational time - dependent hartree method for bosons ( mctdhb ) @xcite , also see refs .
@xcite for a few applications .
the mctdhb many - body wavefunction is a linear combination of all above described _ time - dependent _ permanents @xmath20 the evolutions of the coefficients @xmath21 describing all possible hopping processes in the system and the dynamical changes of the shapes of every fragment @xmath22 are determined by solving the mctdhb equations : @xmath23.\nonumber\end{aligned}\ ] ] here @xmath24 is the single particle hamiltonian , @xmath25 and @xmath26 are the matrix elements of the reduced one- and two - body densities of @xmath27 . the local time - dependent potentials @xmath28 originate from the two - body interaction and play a crucial role in the physics studied here .
to investigate the dynamical stability of a desired state as a time - dependent process we , first , specify this state by providing corresponding initial conditions @xmath29 and @xmath30 , and then monitor how these quantities change in time in a response to the applied modification of the trap @xmath31 and/or to a quench of the interaction strength @xmath2 of @xmath32 .
our goal is to investigate the dynamical stability of the multi - hump multi - fold fragmented states . in ref .
@xcite we have shown that in simple , barrier - less traps the number of humps and fragmentation ratio of the bosonic system with finite- and long - range repulsive interactions in the ground state can be equivalently controlled by varying either the strength of the inter - particle repulsion @xmath2 , the tightness / strength of harmonic confinement @xmath33 , or by changing the number @xmath34 of trapped particles . in the supplemental material @xcite , a detail control of the humps structures of the ground state in parabolic trap and inter - particle interaction @xmath35 of half - width @xmath36 with @xmath37 , studied throughout this work , is presented .
so , from now on we assume that a ground state with a desired humps structures and fragmentation ratio is available . and a simultaneous quench of the inter - particle repulsion .
the initial state is the ground state of @xmath38 bosons confined in @xmath39 with @xmath40 .
minkovski - like space - time evolutions of the density are plotted for the scenarios activated by the displacement of the trap ( a ) without quench of the repulsion ; ( b ) with small increase of the repulsion @xmath41 ; ( c ) with substantial decrease of the repulsion @xmath42 . panels ( a , b ) reveal a first generic regime
non - violent , under - a - barrier dynamics .
panel ( c ) represents a second , over - a - barrier regime with a highly - non - equilibrium , explosive changes of the density .
all quantities shown are dimensionless.,title="fig:",width=94 ] and a simultaneous quench of the inter - particle repulsion .
the initial state is the ground state of @xmath38 bosons confined in @xmath39 with @xmath40 .
minkovski - like space - time evolutions of the density are plotted for the scenarios activated by the displacement of the trap ( a ) without quench of the repulsion ; ( b ) with small increase of the repulsion @xmath41 ; ( c ) with substantial decrease of the repulsion @xmath42 . panels ( a , b ) reveal a first generic regime
non - violent , under - a - barrier dynamics .
panel ( c ) represents a second , over - a - barrier regime with a highly - non - equilibrium , explosive changes of the density .
all quantities shown are dimensionless.,title="fig:",width=94 ] and a simultaneous quench of the inter - particle repulsion .
the initial state is the ground state of @xmath38 bosons confined in @xmath39 with @xmath40 .
minkovski - like space - time evolutions of the density are plotted for the scenarios activated by the displacement of the trap ( a ) without quench of the repulsion ; ( b ) with small increase of the repulsion @xmath41 ; ( c ) with substantial decrease of the repulsion @xmath42 . panels ( a , b ) reveal a first generic regime
non - violent , under - a - barrier dynamics .
panel ( c ) represents a second , over - a - barrier regime with a highly - non - equilibrium , explosive changes of the density .
all quantities shown are dimensionless.,title="fig:",width=94 ] let us first take a two - hump two - fold fragmented system in 1d , obtained as the ground state of @xmath38 bosons confined in @xmath43 with @xmath40 , see fig .
s1(b ) of the supplemental material .
at @xmath44 we suddenly displace the trap with respect to its origin @xmath45 .
the computed density of the evolving many - body wave - packet in a minkovskii - like space - time representation is depicted in fig .
[ fig1](a ) .
the main observation is that this manipulation of the trap induces only a non - violent many - body dynamics the two - hump topology and the two - fold fragmentation of the system persist for all the presented times .
now we enrich the dynamical scenario studied above by imposing onto the same initial system together with the sudden displacement of the trap at @xmath44 also a sudden small quench of the inter - particle repulsion form @xmath41 .
the computed wave - packet evolution is depicted in a minkovski - like manner in fig .
[ fig1](b ) .
the wave - packet dynamics reveals along with a relative simple motion induced by the trap displacement also new additional features .
the sub - clouds forming the wave - packet change their widths during the evolution back and fourth , i.e. , they `` breath '' . these breathings slightly disturb and modulate the perfect harmonic - like oscillations of both sub - clouds .
next , in a third scenario we take the same initial state as before , suddenly displace the trap at @xmath44 but now significantly decrease the strength of the inter - particle repulsion form @xmath42 . at this value of the inter - particle interaction the ground state of the final system
has a one - hump topology and is fully condensed , see fig .
s1(a ) of the supplementary material .
so , this scenario can be considered as an attempt to make a super - fluid from an initially fragmented system .
the corresponding minkovski - like evolution of the density is shown in fig .
[ fig1](c ) .
the wave - packet reveals highly - non - equilibrium explosive dynamics which is accompanied by the formation of complicated oscillating patterns in the density .
the character of this dynamics differs drastically from the evolutions studied above , compare panels ( a - b ) with panel ( c ) of fig .
[ fig1 ] .
the studied 1d many - body systems demonstrate two , generically different reactions to the applied manipulations with the external trapping and inter - particle interaction potentials a non - violent and a highly - non - equilibrium , explosive ones .
a main claim of this letter is that these reactions are generic features of disturbed strongly interacting repulsive systems . to confirm this generality
we have extended the above reported manipulations with trap displacements and quenches of the inter - particle interactions to 2d and 3d setups , see figs .
[ fig2],[fig3 ] and the supplementary material for details . a this point it is worthwhile to stress that all the multi - hump systems studied here are initially fragmented and remain fragmented during the non - violent and explosive time - evolutions . in fragmented systems , there are no definite phase relations ( correlation ) between the sub - fragments , pretty much as in mott - like states @xcite . in fully condensed systems , in contrast
, the phase between different humps is fixed .
fig .
s2 of the supplemental material shows how one - body correlation functions @xcite can be used to distinguish fragmented and condensed systems .
to understand the physics behind the non - violent and explosive regimes of the time - dependent many - body dynamics of the trapped multi - hump multi - fold fragmented systems with strong finite - range inter - particle interactions , we extend the concept of self - induced effective potentials introduced in ref .
@xcite to explain static properties of the multi - hump ground states . as in the static case , we associate each hump with a well - isolated , now time - evolving sub - system @xmath46 . during the propagation , as one can see from the mctdhb equations of motion , eqs .
( [ mctdhb_eq ] ) , each of these fragments `` feels '' the external trap potential and , also , effective time - dependent potential barriers @xmath47 induced by the other ( counterpart ) sub - clouds as a result of the inter - particle interaction .
the intensity , i.e. , the height of the induced , time - dependent barriers , is proportional to the strength of the repulsion @xmath2 and to the ratio @xmath48 of the involved elements of reduced two- and one - body density matrices , see mctdhb equations , eqs .
( [ mctdhb_eq ] ) . in the case of an ideal fragmented state with well - localized fragments ,
this ratio is proportional to the number of particles residing in each fragment .
for instance , in the case of a perfect two - fold fragmented system it approaches @xmath49 , where @xmath34 is the total number of particles . with a simultaneous quench of the inter - particle repulsion : ( a ) strong decrease @xmath42 , snap - shot at @xmath50 ; ( b ) moderate increase @xmath51 , snap - shot at @xmath52 ; ( c ) stronger increase @xmath53 , snap - shot at @xmath54 .
upper panels show the densities @xmath55 of the @xmath56 fragments [ working mctdhb orbitals , eqs .
( [ mctdhb_eq ] ) ] and the time - dependent barriers @xmath57 induced by the complimentary @xmath58 sub - clouds .
trapping potentials and computed density are shown on the lower panels .
the over - a - barrier dynamics ( a , c ) happens when the energy per particle of the out - of - equilibrium state is larger than the heights of the induced barriers , otherwise the dynamics is under - a - barrier ( b ) .
the induced barriers depicted in ( a ) have been multiplied by a factor of 40 , for better visualization .
all quantities shown are dimensionless.,title="fig:",width=217 ] with a simultaneous quench of the inter - particle repulsion : ( a ) strong decrease @xmath42 , snap - shot at @xmath50 ; ( b ) moderate increase @xmath51 , snap - shot at @xmath52 ; ( c ) stronger increase @xmath53 , snap - shot at @xmath54 .
upper panels show the densities @xmath55 of the @xmath56 fragments [ working mctdhb orbitals , eqs .
( [ mctdhb_eq ] ) ] and the time - dependent barriers @xmath57 induced by the complimentary @xmath58 sub - clouds . trapping potentials and computed density are shown on the lower panels .
the over - a - barrier dynamics ( a , c ) happens when the energy per particle of the out - of - equilibrium state is larger than the heights of the induced barriers , otherwise the dynamics is under - a - barrier ( b ) .
the induced barriers depicted in ( a ) have been multiplied by a factor of 40 , for better visualization .
all quantities shown are dimensionless.,title="fig:",width=217 ] with a simultaneous quench of the inter - particle repulsion : ( a ) strong decrease @xmath42 , snap - shot at @xmath50 ; ( b ) moderate increase @xmath51 , snap - shot at @xmath52 ; ( c ) stronger increase @xmath53 , snap - shot at @xmath54 .
upper panels show the densities @xmath55 of the @xmath56 fragments [ working mctdhb orbitals , eqs .
( [ mctdhb_eq ] ) ] and the time - dependent barriers @xmath57 induced by the complimentary @xmath58 sub - clouds . trapping potentials and computed density are shown on the lower panels .
the over - a - barrier dynamics ( a , c ) happens when the energy per particle of the out - of - equilibrium state is larger than the heights of the induced barriers , otherwise the dynamics is under - a - barrier ( b ) .
the induced barriers depicted in ( a ) have been multiplied by a factor of 40 , for better visualization .
all quantities shown are dimensionless.,title="fig:",width=217 ] for illustrative purposes we examine and validate the applicability of this analysis in 2d setups where the dynamics is induced by a sudden displacement of the trap potential @xmath59 with simultaneous quenches of the repulsion .
the initial state is the two - hump two - fold fragmented ground state , trapped in @xmath60 with @xmath40 . in the lower panels of fig .
[ fig2 ] we depict snapshots of the evolving densities and respective trapping potential at several different time - slices for the above described scenarios of the trap displacement and quenches of the repulsion : from @xmath42 , at @xmath61 in the fig . [
fig2](a ) ; from @xmath62 , at @xmath63 in the fig .
[ fig2](b ) ; from @xmath64 , at @xmath65 in the fig .
[ fig2](c ) .
the initial two - hump state studied here has a dominant ( @xmath66 ) contribution from the @xmath67 configuration , indicating on an essentially perfect two - fold fragmentation of the system . during the propagation
the fragmentation ratios change , of course , but the systems still remain two - fold fragmented . in the upper panels of fig .
[ fig2 ] we plot snap - shots of the densities @xmath68 of the @xmath56 left / right sub - clouds and the effective time - dependent barriers @xmath69 induced by the counterpart @xmath58 ( right / left ) fragments at the same time - slices . to make the induced barriers visible in the case of a strong decrease of the interaction from @xmath42 depicted in the fig .
[ fig2](a ) , we have scaled them by a factor of 40 .
the full movies of the respective many - body dynamics are available in the supplemental material @xcite . in fig .
[ fig2 ] one can clearly distinguish two qualitatively different regimes of evolutions . a non - violent one , plotted in the middle ( b ) sub - figure , can be contrasted with highly - non - equilibrium ones , depicted in the left ( a ) and right ( c ) sub - figures . to explore and confirm the existence of these two generic regimes observed in one- and two- dimensional setups also in 3d , we plot in fig .
[ fig3 ] an example of a highly non - equilibrium violent dynamics induced by a sudden displacement of the trap @xmath70 and strong decrease of the finite - range repulsion from @xmath42 . here , to visualize the 3d functions we plot several isosurfaces of the density and an equipotential cut of the trap .
the snap - shots of the densities are taken at @xmath71 .
full movies of this and two other 3d scenarios are provided in the supplemental material @xcite .
let us digest the physics of the above observed dynamical regimes . in the non - violent evolutions
, see e.g. fig .
[ fig1](a , b ) for 1d and fig .
[ fig2](b ) for 2d , the strong repulsion prevents an exchange of the particles between the sub - clouds , so they do not come to a proximity for a contact .
the superposition of the induced time - dependent barriers and external trap results in effective potentials which are high enough to confine the sub - clouds even in the case when they are moving .
hence , the non - violent dynamics of the trapped multi - hump , multi - fold fragmented repulsive systems appears when the manipulations exerted on the system are not strong enough to destroy or overcome the induced time - dependent barriers .
so , the physics behind is an under - a - barrier dynamics .
complimentary , the violent dynamics appears when the induced barriers are not high enough to keep the sub - clouds apart from each other .
indeed , in the above studied sudden decreases of the inter - particle repulsion from @xmath42 , depicted in the fig . [
fig1](c ) for 1d , fig .
[ fig2](a ) for 2d , and fig .
[ fig3 ] for 3d , reduce the heights of the induced barriers , which are proportional to @xmath2 .
the sub - clouds start to leak out and eventually become delocalized over the entire trap .
the @xmath53 quench in 2d , depicted in the fig .
[ fig2](c ) , formally leads to an increase of the heights of the induced barriers , but , simultaneously , it pumps too much internal energy into the system . at this new value of the inter - particle interaction strength the ground state is three - fold fragmented and , hence , the shapes of the initial two - hump sub - clouds are no more optimal .
it also implies that the energy per - particle of each sub - cloud is larger than the heights of the induced time - dependent barriers . during the violent evolutions
the available energy is redistributed between the excited states which have multi - node structures and can be delocalized over the entire trap . as a result
, the density reveals the observed highly violent explosive behavior .
so , the physics behind is a complex over - a - barrier dynamics . and strong , sudden decrease of the repulsion from @xmath42 . to visualize the 3d functions we plot several isosurfaces of the density and an equipotential cut of the trap .
the snap - shots of the density are taken to show initial ( @xmath44 ) , coalescing ( @xmath72 ) , penetrating ( @xmath73 ) and highly - excited ( @xmath74 ) momentary instances of the violent over - a - barrier dynamics .
all quantities shown are dimensionless.,title="fig:",width=132 ] and strong , sudden decrease of the repulsion from @xmath42 . to visualize the 3d functions we plot several isosurfaces of the density and an equipotential cut of the trap .
the snap - shots of the density are taken to show initial ( @xmath44 ) , coalescing ( @xmath72 ) , penetrating ( @xmath73 ) and highly - excited ( @xmath74 ) momentary instances of the violent over - a - barrier dynamics .
all quantities shown are dimensionless.,title="fig:",width=132 ] and strong , sudden decrease of the repulsion from @xmath42 . to visualize the 3d functions we plot several isosurfaces of the density and an equipotential cut of the trap .
the snap - shots of the density are taken to show initial ( @xmath44 ) , coalescing ( @xmath72 ) , penetrating ( @xmath73 ) and highly - excited ( @xmath74 ) momentary instances of the violent over - a - barrier dynamics .
all quantities shown are dimensionless.,title="fig:",width=132 ] and strong , sudden decrease of the repulsion from @xmath42 . to visualize the 3d functions we plot several isosurfaces of the density and an equipotential cut of the trap .
the snap - shots of the density are taken to show initial ( @xmath44 ) , coalescing ( @xmath72 ) , penetrating ( @xmath73 ) and highly - excited ( @xmath74 ) momentary instances of the violent over - a - barrier dynamics .
all quantities shown are dimensionless.,title="fig:",width=132 ] now , we are able to deduce a set of practical recommendations on possible experimental preparations and manipulations of the multi - hump , multi - fold fragmented states .
( i ) once prepared , these states remain very stable and robust with respect to possible imperfectness of experimental setups .
( ii ) a protocol where the non - interacting system is first prepared and then the interaction is diabatically quenched seems to be ineffective because , it would lead to explosive dynamics . (
iii ) finally , a formation of the systems with a desired number of humps ( fragments ) can be provoked and controlled by pre - imposing a weak optical lattice of the required periodicity from the beginning of the quench process . by switching it off afterwords , one
could , in principle , induce only a non - destructive non - violent under - a - barrier dynamics .
concluding , we have shown that the physics behind the two hitherto found generic regimes of the many - body non - equilibrium dynamics of trapped ultracold bose systems with strong repulsive finite / long range interactions is driven by the mechanism of interaction - induced time - dependent barriers .
the non - violent dynamics of the first , under - a - barrier regime is contrasted to a highly - non - equilibrium explosive quantum many - body dynamics of the second , over - a - barrier regime .
the generality of the discovered time - dependent physics is verified in one- , two- , and three - spatial dimensions .
computation time on the bwgrid , hlrs and k100 clusters are greatly acknowledged . a partial financial support by the dfg is acknowledged
. 64 m. girardeau , j. math . phys . * 1 * , 516 ( 1960 ) .
t. kinoshita , t. wenger , and d. s. weiss , phys .
rev .
lett .
* 95 * , 190406 ( 2005 ) .
m. a. baranov , phys .
rep .
* 464 * , 71 ( 2008 ) .
t. lahaye _
et al . _ , rep .
. phys . * 72 * , 126401 ( 2009 ) .
d. comparat , and p. pillet , _ j. opt .
soc . am .
b _ * 6 * , a208 ( 2010 )
. phys .
rev .
a * 82 * , 033412 ( 2010 ) .
phys .
rev .
lett . * 104 * , 195302 ( 2010 ) .
m. viteau , m. g. bason , j. radogostowicz , n. malossi , d. ciampini , o. morsch , and e. arimondo , phys .
rev .
lett .
* 107 * , 060402 ( 2011 ) .
see supplemental material at ` http://link.aps.org/supplemental/ ` for discussion on possible experimental control and realization of the predicted phenomena in `` rydberg - dressed '' ultra - cold systems @xcite , and for movies of the reported 2d and 3d mctdhb dynamics .
a. i. streltsov , phys .
rev .
a * 88 * , 041602(r ) ( 2013 ) .
a. i. streltsov , o. e. alon , and l. s. cederbaum , phys .
rev .
lett .
* 99 , * 030402 ( 2007 ) .
o. e. alon , a. i. streltsov , and l. s. cederbaum , phys .
rev .
a * 77 , * 033613 ( 2008 ) .
a. i. streltsov , o. e. alon , and l. s. cederbaum , phys .
rev .
lett . * 100 * , 130401 ( 2008 ) .
k. sakmann , a. i. streltsov , o. e. alon , and l. s. cederbaum , phys .
rev .
lett .
* 103 * , 220601 ( 2009 ) .
a. i. streltsov , o. e. alon , and l. s. cederbaum , phys .
rev .
lett . * 106 * , 240401 ( 2011 ) .
a. u. j. lode , k. sakmann , o. e. alon , l. s. cederbaum , and a. i. streltsov , phys .
rev .
a * 86 * , 063606 ( 2012 ) .
a. u. j. lode , a. i. streltsov , k. sakmann , o. e. alon , and l. s. cederbaum , proc .
natl .
acad .
sci .
usa * 109 * , 13521 ( 2012 ) .
m. greiner _
et al . _ , nature ( london ) * 415 * , 39 - 44 ( 2002 ) .
k. sakmann , a. i. streltsov , o. e. alon , and l. s. cederbaum , phys .
rev .
a * 78 * , 023615 ( 2008 ) .
# 1@xmath75 *o *
|
here the parameter selection used in this study is discussed .
no attempt has been made to determine precisely values such for the best fit to pipe flow .
the models are not sufficiently quantitative that exact comparisons are called for at this time .
moreover , the phenomena presented in the paper are very robust and for some parameters there simply is not a strong criterion to use to select precise values . the goal is to provide justification for the values used in the paper as well as insight into how the parameters control the dynamics of the models
. only the two rates @xmath24 and @xmath26 need to be determined .
the value of @xmath93 has little impact on the dynamics and has simply been fixed at @xmath94 .
these parameters are determined by fitting to a typical puff from dns as shown in fig .
[ fig : compare_epsilons ] . and
@xmath26 chosen to match a typical puff from dns at @xmath95 . in the left plot @xmath28 while @xmath24 has values 0.02 ( red ) , 0.04 ( blue ) , and 0.06 ( green ) . in the right plot @xmath27 and @xmath26
has values 0.1 ( red ) , 0.2 ( blue ) , and 0.4 ( green ) .
dns is the irregular black curve .
the unlabeled ( blue ) curves in the two plots correspond to the values of @xmath24 and @xmath26 used in the paper .
, width=288 ] the left plot shows the spatial profile of model puffs for three values of @xmath24 , the parameter controlling the final relaxation to parabolic flow .
it is straightforward to select a reasonable value of @xmath24 from such a plot .
note , however , that a scaling of model length scale has been performed to plot model and dns profiles on the same graph ( model lengths have been multiplied by 0.225 ) .
this scaling of length is such that the sharp upstream edge of the model puff occurs over the same distance as in dns .
the upstream edge is largely set by @xmath26 .
( if the scaling of space units between model and dns where known for other reasons , then the spatial profile alone could be used to determine both @xmath24 and @xmath26 . ) the right plot is used then to complete the determination . here
model puffs for different values of @xmath26 are plotted in the @xmath1-@xmath0 plane .
the sharp upstream edge of a puff is the trajectory rising from parabolic flow at @xmath23 , @xmath22 and this is strongly affected by the value of @xmath26 . if @xmath26 is too small then the trajectory is too step ( @xmath1 does not respond quickly enough ) .
if @xmath26 is too large , then @xmath0 does not reach a sufficiently large value .
the values of @xmath27 and @xmath28 chosen for the simulations presented in the paper were arrived at by varying the two values to get the best overall agreement in the spatial profile and phase portrait . as stated in the paper , the two rates @xmath24 and @xmath26
are taken to have the same values as in the continuous model .
this is quite reasonable given the relationship between eqs . ( 2 ) and ( 4 ) .
this leaves choosing the parameters @xmath55 , @xmath58 , and @xmath49 of the map @xmath43 , and the parameters @xmath97 and @xmath98 .
the role of each of these is discussed below . as each parameter
is varied in the following , the remaining parameters take the fixed values used in the paper : @xmath27 , @xmath28 , @xmath57 , @xmath62 , @xmath99 , @xmath60 , and @xmath61 . _
parameter @xmath55 _ : the parameter @xmath55 effectively dictates how many iterates of the map @xmath44 are used per time step of the model .
the effect of the parameter @xmath55 is shown in fig .
[ fig : compare_ks ] where puff solutions are shown in the @xmath1-@xmath0 phase plane for @xmath100 and @xmath57 .
model turbulence is more erratic for @xmath57 than @xmath100 .
when @xmath100 , puff splitting and the transition to sustained turbulence occurs at a smaller value of @xmath52 , but the fact that @xmath52 is smaller on the top row of fig .
[ fig : compare_ks ] only partially accounts for the difference between the top and bottom rows of fig .
[ fig : compare_ks ] . .
top row @xmath100 and bottom row @xmath57 . in each case snapshots
are plotted in the @xmath1-@xmath0 phase plane .
the left two plots show randomly chosen snapshots of solutions with three closely spaced puffs .
the right - most plots are of solutions following a quench from high @xmath52 and contain a large number of puffs .
top row is at @xmath101 and the bottom row is at @xmath102 .
both values of @xmath52 are close to the transition to sustained turbulence for the corresponding value of @xmath55 .
, width=324 ] in addition to this visual comparison , there is the fact that the average slope for a unimodal map with stable chaos is limited to @xmath103 and this is an artificial constraint on the dynamics that comes about from considering one - dimensional dynamics .
( in the case of transient chaos the mean slope can exceed 2 , but there is still a constraint relating the escape rate to the mean slope . )
taking @xmath104 is equivalent to considering multimodal maps and removes the artificial constraint .
note that the model shows puffs , puff splitting and slugs even for @xmath100 .
these features are robust .
however , the additional freedom in the chaotic dynamics by allowing @xmath57 permits the model to achieve a better representation of turbulent flow .
i have not found that using @xmath105 offers noticeable further improvement . _
parameter @xmath58 _
: this parameter controls the leakage rate from the chaotic region of the map .
figure [ fig : compare_betas ] shows examples of states for @xmath63 and @xmath106 . with @xmath58 sufficiently negative , as for @xmath63 in the top row of fig .
[ fig : compare_betas ] , a transition from puffs to slugs occurs that is essentially just a noisy version of the continuous models shown in fig . 2 of the paper .
note that the chaotic wedge first touches the @xmath1-nullcline at @xmath107 and the transition from puffs to slugs occurs near to , but not exactly at , this value . if there are any splitting events they are very rare .
with @xmath64 , as for @xmath106 in the bottom row of fig .
[ fig : compare_betas ] , the transition from puffs to slugs is mediated by splitting events . however , the splitting events are too rare for the model to realistically correspond to pipe flow . as illustrated in fig . [
fig : compare_betas_prime ] , with @xmath108 , model puff splitting is similar to pipe flow .
there appears to be no strong basis to select any particular value of @xmath58 based on a visual examination of the onset of splitting .
the value @xmath109 used in the paper is simply a representative value . .
top row @xmath63 and bottom row @xmath106 . in each case snapshots of solutions
are shown on either side on the transition from localized puffs to expanding turbulence .
upper left : @xmath110 .
upper right : @xmath111 .
lower left : @xmath101 .
lower right : @xmath112 .
, width=324 ] _ parameter @xmath49 _ : this parameter controls the monotone decay of turbulence @xmath0 following exit from the chaotic region .
figure [ fig : compare_gammas ] shows puffs plotted in the @xmath1-@xmath0 phase plane for two value of @xmath49 . for comparison the puff at @xmath95 from fig .
[ fig : compare_epsilons ] is repeated here .
the smaller @xmath49 , the more quickly @xmath0 decays and the less rounded the phase - space dynamics is in the lower left corner of the phase portraits .
the best match of turbulent decay in the model will be at the largest possible value for @xmath49 .
however , as @xmath49 approaches 1 laminar flow becomes marginally stable to @xmath0 perturbations and this is clearly unphysical .
the value @xmath99 was chosen as a compromise between the competing requirements of having @xmath49 large but not too close to 1 .
an improvement could be likely be obtained by having @xmath49 be a function of @xmath1 and @xmath52 , but this introduces additional fitting parameters and is not done here . .
left shows @xmath117 , with @xmath118 , middle shows @xmath59 , with @xmath102 , and is the same plot as bottom middle of fig . 2 .
arrows indicate the relevant region of the @xmath1-@xmath0 phase plane .
the right - most plot is dns of a puff at @xmath95 , exactly the same as in fig . 1 . , width=324 ] _ parameters @xmath97 and @xmath98 _ : these parameters are naturally thought of as arising from the discretization of the terms @xmath17 and @xmath16 in the continuous model . at present
i am not aware of any compelling reasons to select @xmath97 and @xmath98 to particular values other than that they should be small ( @xmath98 must satisfy @xmath119 for stability reasons ) .
the value of @xmath97 was chosen to be less slightly less than 1/2 .
the value of @xmath98 is then to be fixed .
based on the analogy with the continuous model , and the value chosen for @xmath97 , it could be taken to be @xmath120 .
this is because discretizing the continuous model with a time step of @xmath121 means that @xmath122 , from an upwind discretization of the advection term .
then @xmath98 will be @xmath123 from discretization of the diffusion term .
this would give @xmath124 .
however , adjusting @xmath98 downwards from this value to @xmath61 places the transition to sustained turbulence in the model at the critical reynolds number for pipe flow . having the model match
this transition point seems preferable to setting it to the particular value @xmath125 .
moreover , it is common to vary the diffusive coupling constant in studies of coupled - map lattices . to emphasize that qualitative features of the model do not strongly dependent on the parameters @xmath97 and @xmath98 ( as long as they are reasonably small ) , fig .
[ fig : compare_cd ] shows some splitting puffs for different values of @xmath97 and @xmath98 .
the reason for focusing on splitting puffs is these best show the fidelity of the model . puffs and slugs are easily obtained . in each case
a puff was generated and @xmath52 was increased slowly until a splitting occurred .
the value of @xmath52 are given in the caption . and
@xmath98 . in each case
snapshots of solutions are shown just after a puff splitting .
splitting is not very sensitive to @xmath97 and @xmath98 around values used in the paper .
upper left : @xmath60 , @xmath126 , @xmath127 .
upper right : @xmath60 , @xmath128 , @xmath129 .
lower left : @xmath130 , @xmath61 , @xmath131 .
lower right : @xmath132 , @xmath61 , @xmath133 . ,
width=324 ] for three puffs illustrating abrupt decay at unpredictable times ( @xmath112 ) .
bottom row shows exponential ( memoryless ) probabilities @xmath134 for ( b ) decaying puffs ( @xmath135 ) and ( c ) splitting puffs ( @xmath136 ) . , width=324 ] figure [ fig : lifetimes](a ) shows time series of total energy @xmath137 for puffs from three initial conditions in which @xmath0 is varied at one space point by less than @xmath138 . while the dynamics are deterministic , abrupt decay occurs at unpredictable times . from a large number of such simulations lifetime statistics
can be generated .
specifically , simulations were initiated from small perturbations designed to trigger puffs .
a random number generator was used to introduce a small amount of randomness into each perturbation .
simulations were then run for 1000 time steps to allow puffs to equilibrate . if puffs had not decayed , the puffs were used as initial conditions for simulations for decay statistics .
figure [ fig : lifetimes](b ) shows representative survival probabilities @xmath134 of a puff lasting at least time @xmath42 .
each distribution corresponds to 4000 realizations .
the survival functions are exponential ( memoryless ) , @xmath84 , where @xmath85 is the @xmath52-dependent mean lifetime until decay .
puff splitting is similar .
initial conditions were generated in the same way only here equilibration simulations were run for 1400 time steps because at the largest @xmath52 in the study 1000 time steps was not quite enough to remove all equilibration effects .
a puff was defined to have split once two turbulent peaks are separated by least 80 grid points .
figure [ fig : lifetimes](c ) shows representative survival probabilities @xmath134 for a puff to last at least time @xmath42 without splitting .
the distributions are again exponential , @xmath84 , showing that model splitting is indeed memoryless with a mean splitting time @xmath85 . these lifetimes are plotted in fig .
4(a ) of the paper .
the turbulence fraction @xmath72 serves as the order parameter for the onset of sustained turbulence .
a point is defined to be turbulent if @xmath139 where @xmath140 sets a threshold relative to the lower boundary separating chaotic and monotonic decay of the map @xmath43 .
@xmath72 is the mean fraction of grid points in the turbulent state .
means have been computed from 4 independent simulations .
for @xmath52 near the critical point , simulations of @xmath141 time steps were run on grids of @xmath142 points .
the standard deviation from the 4 independent simulations is comparable to the point size in fig .
4(b ) of the paper .
for @xmath143 simulations for @xmath144 time steps on grids of @xmath145 points were more than sufficient . in fig . 4 of the paper ,
@xmath72 is plotted for @xmath146 . while the exact value of @xmath72 depends on threshold @xmath140 , the extent of the intermittent region and critical scaling at onset do not .
37ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty http://www.jstor.org/stable/109431 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevlett.91.244502 [ * * , ( ) ] link:\doibase 10.1017/s0022112007006398 [ * * , ( ) ] link:\doibase 10.1007/bf00536526 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1073/pnas.0909560107 [ * * , ( ) ] link:\doibase 10.1017/s0022112004008134 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.96.094501 [ * * , ( ) ] link:\doibase 10.1038/nature05089 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.98.014501 [ * * , ( ) ] link:\doibase 10.1103/physreve.78.046310 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.101.214501 [ * * , ( ) ] link:\doibase 10.1017/s0022112009993296 [ * * , ( ) ] link:\doibase 10.1017/s0022112009993065 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1017/s0022112008003315 [ * * , ( ) ] link:\doibase 10.1017/s0022112010003435 [ * * , ( ) ] @noop @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physreve.79.025301 [ * * , ( ) ] link:\doibase 10.1016/j.jcp.2004.02.013 [ * * , ( ) ] link:\doibase 10.1126/science.1186091 [ * * , ( ) ] http://books.google.com / books?hl = en&lr=&id = qvuyro8izcec&oi = fnd&pg = pa221&dq% = % 2522sreenivasan%2522+%2522relaminarization%2522+&ots = sftiztux4d&sig = rhtktphu% _ udxqmm3qagkejtx8qi [ * * , ( ) ] @noop * * , ( ) @noop _ _ , ed .
( , , ) @noop * * , ( ) link:\doibase 10.1146/annurev.fluid.39.050905.110308 [ * * , ( ) ] link:\doibase 10.1088/1367 - 2630/11/1/013040 [ * * , ( ) ] link:\doibase 10.1103/physreve.81.035304 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.99.034502 [ * * , ( ) ] link:\doibase 10.1098/rsta.2008.0216 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.103.054502 [ * * , ( ) ] @noop * * , ( )
|
transitional pipe flow is modeled as a one - dimensional excitable and bistable medium .
models are presented in two variables , turbulence intensity and mean shear , that evolve according to established properties of transitional turbulence .
a continuous model captures the essence of the puff - slug transition as a change from excitability to bistability . a discrete model , that additionally incorporates turbulence locally as a chaotic repeller , reproduces almost all large - scale features of transitional pipe flow .
in particular it captures metastable localized puffs , puff splitting , slugs , a continuous transition to sustained turbulence via spatiotemporal intermittency ( directed percolation ) , and a subsequent increase in turbulence fraction towards uniform , featureless turbulence .
the transition to turbulence in pipe flow has been the subject of study for over 100 years @xcite , both because of its fundamental role in fluid mechanics and because of the detrimental consequences of turbulent transition in many practical situations .
there are at least two features of the problem that make it fascinating , but also difficult to analyze .
the first is that when turbulence appears , it appears abruptly @xcite , and not through a sequence of transitions each increasing the dynamical complexity of the flow .
turbulence is triggered by finite - sized disturbances to linearly stable laminar flow @xcite .
this hysteretic , or subcritical , aspect of the problem limits the applicability of linear and weakly nonlinear theories .
the second complicating feature is the intermittent form turbulence takes in the transitional regime near the minimum reynolds number ( non - dimensional flow rate ) for which turbulence is observed @xcite . in sufficiently long pipes , localized patches of turbulence ( puffs ) may persist for extremely long times before abruptly reverting to laminar flow @xcite .
in other cases , turbulent patches may spread by contaminating nearby laminar flow ( puff splitting and slugs ) @xcite .
while minimal models have been very useful in understanding generic features of intermittency in subcritical shear flows @xcite , such models do not capture the puffs , puff splitting , and slugs that are essential to the character of pipe flow . in this paper
i argue that transitional pipe flow should be viewed in the context of excitable and bistable media . with this perspective
i present models , based on the interaction between turbulence and the mean shear , that both capture and organize most large - scale features of transitional pipe flow .
degrees of freedom in a periodic pipe 200d long .
flow is from left to right . shown
are instantaneous values of turbulence intensity @xmath0 and axial velocity @xmath1 along the pipe axis .
( a ) equilibrium puff at @xmath2 .
( b ) puff splitting at @xmath3 .
the downstream ( right ) puff split from the upstream one at an earlier time .
( c ) expanding slug flow at @xmath4 .
right column shows corresponding states from the simple one - dimensional model - ( d ) @xmath5 , ( e ) @xmath6 , and ( f ) @xmath7 .
, width=324 ] figure [ fig : dns_model ] summarizes the three important dynamical regimes of transitional pipe flow .
the left column shows results from direct numerical simulations ( dns ) @xcite .
quantities are nondimensionalized by the pipe diameter @xmath8 and the mean ( bulk ) velocity @xmath9 .
the reynolds numbers is @xmath10 , where @xmath11 is kinematic viscosity .
flows are well represented by two quantities , the turbulence intensity @xmath0 and the axial ( streamwise ) velocity @xmath1 , sampled on the pipe axis . specifically , @xmath0 is the magnitude of transverse fluid velocity ( scaled up by a factor of 6 ) .
the centerline velocity @xmath1 is relative to the mean velocity and is a proxy for the state of the mean shear that conveniently lies between 0 and 1 . at low @xmath12 , as in fig .
[ fig : dns_model](a ) , turbulence occurs in localized patches propagating downstream with nearly constant shape and speed .
these are called equilibrium puffs @xcite , a misnomer since at low @xmath12 puffs are only metastable and eventually revert to laminar flow , i.e. decay @xcite .
asymptotically the flow will be laminar parabolic flow , @xmath13 , throughout the pipe . for intermediate @xmath12 , as in fig .
[ fig : dns_model](b ) , puff splitting frequently occurs @xcite . new puffs
are spontaneously generated downstream from existing ones and the resulting pairs move downstream with approximately fixed separation .
further splittings will occur and interactions will lead asymptotically to a highly intermittent mixture of turbulent and laminar flow @xcite . at
yet higher @xmath12 , turbulence is no longer confined to localized patches , but spreads aggressively in so - called slug flow @xcite , as illustrated in fig .
[ fig : dns_model](c ) .
the asymptotic state is uniform , featureless turbulence throughout the pipe @xcite .
models for these dynamics will be based on the following known physical features . at the upstream ( left in fig .
[ fig : dns_model ] ) edge of turbulent patches , laminar flow abruptly becomes turbulent .
energy from the laminar shear is rapidly converted into turbulent motion and this results in a rapid change to the mean shear profile @xcite . in the case of puffs , the turbulent profile is not able to sustain turbulence and thus there is a reverse transition @xcite from turbulent to laminar flow on the downstream side of a puff . in the case of slugs ,
the turbulent shear profile can sustain turbulence indefinitely ; there is no reverse transition and slugs grow to arbitrary streamwise length @xcite . on the downstream side of turbulent patches
the mean shear profile recovers slowly @xcite , seen in the behavior of @xmath1 in fig .
1 . crucially
, the degree of recovery dictates how susceptible the flow is to re - excitation into turbulence @xcite .
these are the characteristics of excitable and bistable media @xcite .
in fact the puff in fig .
[ fig : dns_model](a ) bears a close resemblance to an action potential in a nerve axon @xcite .
linearly stable parabolic flow is the excitable rest state , turbulence is the excited state , and the mean shear is the recovery variable controlling the threshold for excitation .
thus , i propose to model pipe flow as a generic excitable and bistable medium incorporating the minimum requisite features of pipe turbulence .
the models are expressed in variables @xmath0 and @xmath1 depending on distance along the pipe .
consider first the continuous model @xmath14 where @xmath15 plays the role of @xmath12 .
@xmath9 accounts for downstream advection by the mean velocity , and is otherwise dynamically irrelevant since it can be removed by a change of reference frame .
the model includes minimum derivatives , @xmath16 and @xmath17 , needed for turbulent regions to excite adjacent laminar ones and for left - right symmetry breaking .
the core of the model is seen in the @xmath0-@xmath1 phase plane in fig .
[ fig : pde ] .
the trajectories are organized by the nullclines : curve where @xmath18 and @xmath19 for the local dynamics ( @xmath20 ) . for all @xmath15 the nullclines intersect in a stable , but excitable , fixed point corresponding to laminar parabolic flow .
the @xmath1 dynamics with @xmath21 captures in the simplest way the behavior of the mean shear . in the absence of turbulence ( @xmath22 )
, @xmath1 relaxes to @xmath23 at rate @xmath24 , while in response to turbulence ( @xmath25 ) , @xmath1 decreases at a faster rate dominated by @xmath26 .
values @xmath27 and @xmath28 give reasonable agreement with pipe flow .
( see the appendix sec .
[ sec : continuous ] . )
the @xmath0-nullcline consists of @xmath22 ( turbulence is not spontaneously generated from laminar flow ) together with a parabolic curve whose nose varies with @xmath15 , while maintaining a fixed intersection with @xmath22 at @xmath29 , ( @xmath30 is used here ) .
the upper branch is attractive , while the lower branch is repelling and sets the nonlinear stability threshold for laminar flow .
if laminar flow is perturbed beyond the threshold ( which decreases with @xmath15 like @xmath31 ) , @xmath0 is nonlinearly amplified and @xmath1 decreases in response .
the ( excitable ) puff regime occurs for @xmath32 , figs .
[ fig : pde](a ) and ( c ) .
the upstream side of a puff is a trigger front @xcite where abrupt laminar to turbulent transition takes place .
however , turbulence can not be maintained locally following the drop in the mean shear .
the system relaminarizes ( reverse transition ) on the downstream side in a phase front @xcite whose speed is set by the upstream front .
following relaminarization , @xmath1 relaxes and laminar flow regains susceptibility to turbulent perturbations .
the slug regime occurs for @xmath33 , figs .
[ fig : pde](b ) and ( d ) . the nullclines intersect in additional fixed points .
the system is bistable and turbulence can be maintained indefinitely in the presence of modified shear .
both the upstream and downstream sides are trigger fronts , moving at different speeds , giving rise to an expansion of turbulence .
a full analysis will be presented elsewhere . and
( b ) @xmath34 .
the fixed point @xmath35 corresponds to parabolic flow . in ( b )
the additional stable fixed point corresponds to stable turbulence .
solution snapshots show ( c ) a puff at @xmath36 and ( d ) a slug at @xmath34 .
these solutions are plotted in the phase planes with arrows indicating increasing @xmath37 .
, width=307 ] while eqs . - capture the basic properties of puffs and slugs , the turbulence model is too simplistic to show puff decay and puff splitting .
evidence suggests that pipe turbulence is locally a chaotic repeller @xcite . hence a more realistic model , within the two - variable excitability setting , is obtained by replacing the upper turbulent branches in fig .
[ fig : pde ] with a wedged - shaped region of transient chaos , illustrated in fig .
[ fig : model ] . outside this region @xmath0 decays monotonically .
the model is @xmath38 where @xmath39 and @xmath40 denote values at spatial location @xmath41 and time @xmath42 .
this model is essentially a discrete version of eqs . - , except with chaotic @xmath0 dynamics generated by the map @xmath43 .
the map @xmath43 is based on models of chaotic repellers in shear flows @xcite .
consider the tent map @xmath44 given by @xmath45 with @xmath46 and @xmath47 .
parameter @xmath48 marks the lower boundary separating chaotic and monotonic dynamics , fig . [ fig : model](b ) , while @xmath49 sets the decay rate to the fixed point @xmath22 . for @xmath50 ( @xmath51
the map generates transient ( persistent ) chaos within the tent region . the map is incorporated into the pipe model by having the threshold @xmath48 depend on @xmath1 as well as on a control parameter @xmath52 , via @xmath53 the factor @xmath54 generates the desired wedged - shaped region , while 2000 sets the scale of @xmath52 to that of @xmath12 .
finally , the map @xmath43 is given by @xmath55 iterates of @xmath44 , i.e. @xmath56 ; with @xmath57 used here .
( see the appendix sec .
[ sec : discrete ] . )
this has the effect of increasing the lyapunov exponent within the chaotic region .
-@xmath0 phase plane . within a wedge - shaped region @xmath0 undergoes transient chaos , while outside it decays monotonically to @xmath22 .
the region varies with @xmath52 as indicated .
( b ) map used to produce transient chaos .
parameter @xmath48 ( which depends on @xmath1 and @xmath52 ) , is the lower boundary separating monotonic and chaotic dynamics .
, width=324 ] the only important new parameter introduced in the discrete model is @xmath58 since it quantifies a new effect spontaneous decay of local turbulence for @xmath50 .
suitable values for others are : @xmath27 and @xmath28 as before , @xmath59 , @xmath60 and @xmath61 .
( see the appendix sec .
[ sec : discrete ] . ) as shown in fig .
[ fig : dns_model ] , for @xmath62 the model shows puffs , puff splitting , and slugs remarkably like those from full dns .
the model parameter @xmath52 nearly corresponds to reynolds number @xmath12 . while positive @xmath58 is ultimately of interest , to better connect the two models
consider first @xmath58 negative , e.g. @xmath63 .
a transition from puffs to slugs occurs as @xmath52 increases and the wedge of chaos crosses the @xmath1-nullcline .
one finds a noisy version of the continuous model in fig .
[ fig : pde ] .
( see the appendix fig .
[ fig : compare_betas ] . )
if splittings of turbulent patches occur , they are exceedingly rare . at @xmath64 , ( including even @xmath65 ) , chaotic fluctuations in @xmath0 cause occasional splitting of expanding turbulence .
puffs at lower @xmath52 are clearly metastable , persisting for long times before decaying . however , splitting and decay are unrealistically infrequent if @xmath58 is too small .
setting @xmath66 gives realistic behavior , as seen in fig .
[ fig : dns_model ] where @xmath62 .
( see also the appendix fig .
[ fig : compare_betas_prime ] . ) note that the splitting of expanding turbulent patches and the decay of localized puffs are caused by the same process - the collective escape from the chaotic region of a sufficiently large streamwise interval to bring about local relaminarization .
this is precisely the scenario described by extreme fluctuations @xcite . in the case of puffs ,
this results in puff decay , while in the case of splitting , laminar gaps open whose sizes are then set by the recovery of the slow @xmath1 field . , ( b ) puff splitting at @xmath67 , and ( c ) slug formation from an edge state at @xmath68 .
for ease of comparison with published work on pipe flow , solutions are shown in a frame co - moving with structures .
turbulence intensity @xmath0 is plotted with @xmath69 in white . in ( a ) and ( b )
the scale is linear with @xmath22 in black , while in ( c ) the scale is logarithmic with @xmath70 in black .
dimension bars indicate space and time scales .
the top space scale applies also to ( b ) .
, width=319 ] figure [ fig : visu ] further illustrates how well the discrete model captures the three regimes of transitional pipe flow .
spacetime plots show puff decay , puff splitting , and slug flow . in fig .
[ fig : visu](a ) , a puff persist for only a finite time before abruptly decaying @xcite . in fig .
[ fig : visu](b ) , puff splitting dominates the dynamics @xcite .
new puffs are generated downstream from existing ones such that intermittent turbulent regions fill space .
compare especially with refs . . finally , in fig .
[ fig : visu](c ) , a slug arises from a localized edge state ( a low - amplitude state on the boundary separating initial conditions which evolve to turbulence from those which decay to laminar flow @xcite ) .
compare especially with ref . .
is plotted throughout the transitional regime .
the onset of sustained turbulence , via spatiotemporal intermittency , occurs continuously at @xmath71 .
@xmath72 increases with @xmath52 and saturates near @xmath73 .
asymptotic regimes ( laminar , intermittent , featureless ) are labeled , along with corresponding transient dynamics ( decaying puffs , puff splitting , slugs ) .
the onset of featureless turbulence is not sharp as indicated by gray shading .
space - time plots illustrate the dynamics near the ends of the transitional regime ( @xmath74 , @xmath75 , @xmath76 ) with @xmath0 plotted in frames co - moving with structures ( color map indicated , dark is laminar ) .
( a ) mean lifetimes for decaying ( circles ) and splitting ( squares ) puffs crossing at @xmath77 . ( b ) log - log plot of @xmath72 versus @xmath78 .
best fit to the filled ( red ) points determines @xmath79 and the slope .
, width=324 ] the remainder of the paper provides a global perspective of the transitional regime , obtained from extensive numerical simulation of eqs . - and summarized in fig .
[ fig : ft ]
. turbulence fraction @xmath72 serves as the order parameter , tracking the dynamics from the onset of intermittency through the approach to uniform , featureless turbulence .
a point is defined to be turbulent if @xmath80 and @xmath72 is the mean fraction of turbulent points .
there is a continuous transition to sustained turbulence , via spatiotemporal intermittency @xcite , at a critical value @xmath81 .
below @xmath79 , the flow is asymptotically laminar and @xmath82 . above @xmath79 turbulence persists indefinitely and @xmath83 .
this transition is associated with the crossing of mean lifetimes for puff decay and splitting shown in fig .
[ fig : ft](a )
. both decay and splitting are memoryless processes with exponential survival distributions @xmath84 , where @xmath85 is the @xmath52-dependent mean time until decay or split .
( see the appendix sec .
[ sec : stats ] and fig .
[ fig : lifetimes ] . )
the mean lifetimes vary approximately super - exponentially with @xmath52 @xcite , but neither is exactly of this form .
mean lifetimes cross at @xmath77 .
above @xmath86 an isolated puff is more likely to split than decay .
as expected , even though individual turbulent domains may still decay , others may split , as seen in the spacetime plot at @xmath74 .
due to correlations between splitting and decay events , the critical value @xmath79 is not identical to @xmath86 , but is very close to it ( a difference of @xmath87 ) . fig .
[ fig : ft](b ) shows that just above criticality , @xmath88 , supporting that the transition falls into the universality class of directed percolation @xcite .
the ratio of turbulence to laminar flow increases through the intermittent region and at the upper end only small laminar flashes are seen within a turbulent background . beyond @xmath89
laminar regions essentially disappear and @xmath90 .
this occurs in pipe flow at @xmath91 @xcite .
the transition to featureless turbulence is not sharp , however , nor is the transition from puff splitting to slugs .
this upper transition will be addressed elsewhere , but the basic effect , common for bistable media , is already contained in eqs . - . for a range of @xmath15 above the slug transition , ( @xmath92 )
, turbulence does not expand to fill the domain in the presence of other slugs .
small laminar regions remain due to the recovery of the slow @xmath1-field and this sets the scale for the laminar flashes at the upper end of the transition region in fig .
[ fig : ft ] .
i have sought to understand key elements of transitional pipe flow
puffs , puff splitting , and slugs without appealing in detail to the underlying structures within shear turbulence .
this approach is similar to that expounded by pomeau @xcite , and considered elsewhere @xcite .
the important insight here is the close connection between subcritical shear flows and excitable systems .
the view is that a great many features of intermittent pipe flow can be understood as a generic consequence of the transition from excitability to bistability where the turbulent branch is itself locally a chaotic repeller .
i have introduced particular model equations to express these ideas in simple form .
while phenomena have been demonstrated with specific parameters , the phenomena are robust .
the challenge for future work is to obtain more quantitatively accurate models , perhaps utilizing full simulations of pipe flow , since ultimately the fluid mechanics of shear turbulence ( streaks and streamwise vortices ) is important for the details of the process .
more challenging is to extend this effort to other subcritical shear flows , such a plane channel , plane couette , and boundary - layer flows .
these require non - trivial extensions of the current work because , unlike here , the mean shear profile can not obviously be well captured by a simple scalar field .
i thank m. avila , y. duguet , b. hof , p. manneville , d. moxey , and l. tuckerman for valuable discussions .
computing resources were provided by idris ( grant 2010 - 1119 ) .
|
it is well known that isospin invariance is not an exact symmetry of strong interactions . consequently , nuclear forces have a small but measurable charge - dependent component . the equality between proton - proton ( @xmath6 ) [ or neutron - neutron ( @xmath7 ) ] and neutron - proton ( @xmath8 ) nuclear interactions
is known as charge independence .
charge - independence breaking ( cib ) is seen most clearly in the @xmath2 nucleon - nucleon ( @xmath0 ) scattering lengths .
the latest empirical values for the singlet scattering length @xmath9 and effective range @xmath10 are @xcite : @xmath11 the values given for @xmath6 and @xmath7 scattering refer to the nuclear part of the interaction as indicated by the superscript @xmath12 .
electromagnetic effects have been removed from the experimental values , which is model dependent .
the uncertainties quoted for @xmath13 and @xmath14 are mainly due to this model dependence .
it is useful to define the following averages : @xmath15 by definition , cib is the difference between the @xmath8 values and these averages : @xmath16 thus , the @xmath0 singlet scattering length shows a clear signature of cib in strong interactions .
charge dependence of @xmath0 interactions has been the subject of extensive investigations , both experimentally and theoretically , for many decades ( for recent reviews , see refs .
the current understanding is that the charge dependence of nuclear forces is due to a differences in the up and down quark masses and electromagnetic interactions . on a more phenomenological level ,
major causes of cib are : * mass splitting of isovector mesons ; particularly , @xmath4 and @xmath17 ; * irreducible pion - photon exchanges . it has been known for a long time that the difference between the charged and neutral pion masses in the one - pion - exchange ( ope ) potential accounts for about 50% of @xmath18 . in ref .
@xcite , charge dependent interactions were derived for _ np _ and _ pp _ scattering , based on a preliminary version of the bonn meson - exchange model @xcite taking into account the pion mass difference in ope as well as two - boson exchanges . with these interactions ,
about 80% of the empirical @xmath18 could be explained .
earlier , ericson and miller @xcite had obtained a very similar result using the meson - exchange model of partovi and lomon @xcite . the calculations of refs .
@xcite were performed only for the singlet scattering length .
however , it is also of interest to know the charge - dependent effects for intermediate energies and in partial waves other than @xmath2 .
therefore , it is the main purpose of the present investigation to calculate phase shift differences between _ pp _ ( or @xmath7 ) and _ np _ scattering for states with total angular momentum j@xmath19 4 and laboratory incident kinetic energies @xmath20 300 mev .
this paper complements an earlier one on charge - asymmetry of the @xmath0 interaction @xcite . in sect .
ii , we will discuss various classes of irreducible meson - exchange diagrams and calculate their cib effects
due to pion - mass splitting on @xmath0 phase shifts and singlet effective range parameters .
summary and conclusions are given in sect .
the bonn meson - exchange model for the @xmath0 interaction has been described in detail in the literature @xcite .
it is a field - theoretic model that , apart from the well - known one - boson - exchange terms , includes an explicit model for the @xmath1-exchange , @xmath21 diagrams , and some further contributions of @xmath22- and @xmath23-exchange .
the bonn model yields an excellent description of the @xmath0 scattering data below pion production threshold @xcite and , thus , provides a reliable basis for an investigation of the charge dependence of the @xmath0 interaction . within the model , charge - dependence
is created by the mass difference between the charge - states of mesons .
the bonn model includes three isovector mesons , namely , @xmath4 , @xmath24 , and @xmath25 .
we will focus here mainly on the charge - dependent effects due to pion mass difference .
effects due to rho - mass splitting will be discussed briefly at the end of this section , and mass splitting of the @xmath25 will be ignored since nothing is known .
we use averages for the baryon masses : the average nucleon mass @xmath26 mev and the average @xmath27-mass @xmath28 mev .
the pion masses are @xmath29 the values are based upon the 1992 review of particle properties @xcite .
the interaction lagrangians involving pions are @xmath30 with @xmath31 the nucleon , @xmath32 the @xmath27 ( rarita - schwinger spinor ) , and @xmath33 the pion fields .
@xmath34 are the usual pauli matrices describing isospin 1/2 and @xmath35 is the isospin transition operator .
denotes the hermitean conjugate .
the lagrangians are devided by @xmath36 to make the coupling constants @xmath37 dimensionless . following established conventions @xcite
, we always use @xmath36 as scaling mass .
it may be tempting to use @xmath38 for @xmath39 coupling .
notice , however , that the scaling mass could be anything .
therefore , it is reasonable to keep the scaling mass constant within su(3 ) multiplets @xcite .
this avoids the creation of unmotivated cib . in our investigation of charge - dependent effects on the @xmath0 interaction ,
we start from a case that may be denoted the average between @xmath6 and @xmath7 scattering . for this case ,
our model yields @xmath40 fm for the singlet scattering length and 2.864 fm for the effective range , consistent with eqs .
( 2 ) and ( 3 ) .
the one - pion - exchange contribution for this average case is depicted in fig .
1a and @xmath1-exchange contributions are shown in fig .
2a , 3a , and 4a .
note that , in this case , the proton ` p ' in figs .
1a to 4a carries the average nucleon mass of 938.919 mev and there are no electromagnetic interactions ; equally well , one may use a neutron ` n ' in place of the proton in part ( a ) of all figures . to calculate the effects of charge dependence on the @xmath0 phase shifts , we introduce for each @xmath41 state the cib phase shift difference @xmath42 , defined by @xmath43 where @xmath44 denotes the average of the @xmath6 and @xmath7 phase shifts which
, as discussed , is calculated by taking the diagrams figs .
1a to 4a into account ( besides the other diagrams involved in the bonn model ) with average nucleon mass and all electromagnetic interactions switched off .
the phase shift @xmath45 is the @xmath8 one to be calculated below .
similarly , we define the cib mixing parameter difference @xmath46 , @xmath47 the charge - dependence generated by the model under consideration is now ` switched on ' step by step : 1 . * one - pion - exchange * ( ope ) , fig . 1 :
the cib effect is created by replacing the diagram of fig .
1a by the two diagrams of fig .
note that one - meson - exchange contributions are roughly proportional to 1/@xmath48 ( with @xmath49 the meson mass ) because this is approximately the momentum - space one - meson propagator for very low momentum transfer .
thus , since the @xmath39 has a smaller mass than the @xmath50 , @xmath39 exchange is stronger than @xmath50 exchange .
for this reason , ope is stronger in @xmath6 as compared to @xmath8 .
since ope is repulsive in @xmath2 , this phase shift becomes more attractive ( i. e. , larger ) when going from @xmath6 to @xmath8 , resulting in a positive @xmath51 ; cf .
column ` ope ' in table i and dashed curve in fig .
consistent with this is the well - known fact that ope takes care of about 50% of @xmath18 . in the other partial waves , the sign of @xmath51 due to ope depends on
if ope is repulsive or attractive ( e. g. , it is repulsive in @xmath52 and attractive in @xmath53 and @xmath5 ) . due to the small mass of the pion
, ope is a sizable contribution in all partial waves including higher partial waves ; and due to the pion s relatively large mass splitting ( 3.4% ) , ope creates relatively large charge - dependent effects in all partial waves ( fig . 5 and table i ) .
* @xmath1-exchange with @xmath0 intermediate states * ( @xmath54 ) , fig 2 : notice first that only non - iterative ( irreducible ) diagrams are to be considered , since the iterative ones are generated by the scattering equation from ope .
we mention here that , in our approach which is based upon time - ordered perturbation theory , we always take all time - orderings into account ( except for those that imply anti - baryons in intermediate states ) ; however , to save space , we display only a few characteristic time - orderings in fig . 2 ( this is also true for all diagrams shown in figs . 3 and 4 ; to get an impression of the total number of time - ordered diagrams , see fig .
20 of ref .
the cib effect is obtained by replacing the diagrams fig .
2a ( @xmath6/@xmath7 scattering ) by those of fig .
2b ( @xmath8 scattering ) .
for a good understanding of cib effects , it is important to distinguish between box ( here : stretched box ) and crossed box diagrams .
concerning the effect from stretched box diagrams , one replaces the left diagram of fig .
2a by the four stretched box diagrams of fig .
2b ( and similarly for the other stretched box time - orderings not shown ) .
notice now that in the former diagram two @xmath39 are exchanged making this a ` strong ' diagram , while the latter four diagrams together with their isospin factors result in a weaker contribution .
since 2@xmath4 exchange is , in general , attractive , there is a loss of attraction when going from @xmath6 to @xmath8 ( equivalent to a reduction of the phase shift ) .
this qualitative estimate is clearly confirmed by the quantitative results displayed in column ` @xmath55 ' of table ii . the cib effect that stems from crossed box diagrams
is obtained by replacing the two crossed boxes of fig .
2a by the three crossed boxes of fig .
2b . typically , this effect ( column ` @xmath56 ' of table ii ) is of opposite sign as compared to the corresponding ( stretched ) box effect , in most partial waves .
the total cib effect from all diagrams of fig .
2 is displayed by the dashed curve in fig .
* @xmath1-exchange with n@xmath27 intermediate states * ( @xmath57 ) , fig . 3 : notice again , that every box stands for all possible time - orderings of the box type ( diagrams 1 to 6 of fig
. 20 of ref .
@xcite ) and every crossed box for all possible time - orderings of the crossed box type ( diagrams 7 to 12 of fig .
20 of ref .
thus , the total number of diagrams which fig .
3a stands for is 24 while fig .
3b stands for 48 diagrams , which are all explicitly taken into account in our calculations .
replacement of the boxes fig .
3a by the boxes fig .
3b causes an increase in the strength of these diagrams which , since these are attractive diagrams , causes an increase in attraction .
column ` @xmath58 ' in table ii clearly confirms this . for the crossed boxes one gets typically the opposite effect ( column ` @xmath59 ' of table ii ) .
this partial cancelation of the effects from the two groups of diagrams is also demonstrated in fig .
6 where the dash - dot curve represents the effect from the box diagrams while the dash - triple - dot curve is from the crossed ones .
notice that the cancelation is almost perfect in @xmath2 and @xmath5 even though the individual contributions are rather large .
* @xmath1-exchange with @xmath60 intermediate states * ( @xmath61 ) , fig . 4 : the replacement of the diagrams of fig .
4a by fig .
4b shows the by now familiar characteristic : opposite effects from box and crossed box diagrams ( column ` @xmath62 ' and ` @xmath63 ' of table ii ) .
this results in large cancelations between effects which , due to the short - range nature of this class of diagrams , are individually already rather small .
this explains why the cib effect from the diagrams of fig .
4 is negligible in most partial waves ( dotted curve in fig .
+ this finishes the discussion of all contributions of the @xmath1 type . in summary
, one can say that the total cib effect from @xmath1 ( dash - dot curve in fig .
5 ) is quite noticable up to the @xmath64-state . in @xmath2 , @xmath53 , @xmath52 , and @xmath65 ,
the cib effect from @xmath1 is 20 - 50% of the one from ope , at 300 mev .
however , for low energies ( except in @xmath2 ) as well as in higher partial waves , the cib @xmath1 effect is negligible .
* @xmath21-exchanges .
* this group is , in principal , as comprehensive as the @xmath1-exchanges discussed above .
graphically , the @xmath21 diagrams can be obtained by replacing in each diagram of figs . 24 , one of the two pions by a @xmath17-meson of the same charge - state ( because of this simple analogy
, we do not show the @xmath21 diagrams explicitly here ) . concerning the @xmath21 diagrams with @xmath27 intermediate states
a comment is in place . in the bonn model @xcite ,
the crossed @xmath21 diagrams with @xmath66 and @xmath60 intermediate states are included in terms of an approximation .
it is assumed that they differ from the corresponding box diagrams only by the isospin factor .
thus , the @xmath21 box diagrams with @xmath66 and @xmath60 intermediate states are multiplied by an isospin factor that is equal to the sum of the isospin factors for box and crossed box . in this approximation
, these diagrams do not generate any cib effects due to pion - mass splitting . since these diagrams are of very short range , their cib effect may be negligible , anyhow .
the only class of @xmath21 diagrams which we include in our calculation of cib effects is the one that corresponds to fig . 2 , with one pion in each diagram replaced by a @xmath17 meson .
its contribution to cib ( column @xmath21 of table i and dash - triple - dot curve of fig .
5 ) is generally small , and typically opposite to the one from @xmath1 , in most states .
* further @xmath22 and @xmath23 contributions * ( @xmath67 ) .
the bonn potential also includes some @xmath22-exchanges that can be approximated in terms of @xmath68 diagrams and @xmath23-exchanges of @xmath69 type .
it was found in ref .
@xcite that the sum of these contributions is small .
these diagrams have @xmath0 intermediate states
similar to fig . 2 , but with one of the two exchanged pions replaced by an isospin - zero boson
; thus , the isospin factors are different from the ones of fig . 2 and , in fact , like the ones of fig . 1 .
another way of creating these diagrams is to combine the diagrams of fig . 1 with a sigma or an omega in an irreducible way , i. e. , by forming a stretched box or crossed box diagram .
these diagrams carry the same isospin factors as ope .
since this class of diagrams is part of the bonn model , we include these diagrams in our cib consideration .
the cib effect from this class is very small , except in @xmath2 , @xmath52 , and @xmath5 ( column ` @xmath70 ' of table i and dotted curve in fig . 5 ) .
this effect has always the same sign as the effect from ope , but it is substantially smaller .
the reason for the ope character of this contribution is that @xmath68 prevails over @xmath69 and , thus , determines the character of this contribution .
since sigma - exchange is negative and since , futhermore , the propagator in between the @xmath4 and the @xmath71 exchange is also negative , the overall sign of the @xmath68 exchange is the same as ope .
thus , it is like a weak , short - ranged ope .
this finishes our detailed presentation of the diagrams and their cib effects included in our calculation .
the sum of all cib effects on phase shifts is given in the last column of table i and plotted by the solid curve in fig .
notice that the difference between the solid curve and the dashed curve ( ope ) in fig .
5 represents the sum of all effects beyond ope .
thus , it is clearly seen that ope dominates the cib effect in all partial waves , even though there are substantial contributions besides ope in some states , notably @xmath2 , @xmath52 , and @xmath65 . finally , in table iii and iv
, we also give the cib contributions to the @xmath2 scattering length and effective range .
note that the relationship between the cib potential and the corresponding change of the scattering length , @xmath18 , is highly non - linear . as discussed in refs .
@xcite , when the scattering length changes from @xmath72 to @xmath73 due to a cib potential @xmath74 , the relationship is @xmath75 or @xmath76 with @xmath77 and @xmath78 the zero - energy @xmath2 wave functions normalized such that @xmath79 .
thus , the perturbation expansion concerns the invers scattering length . as clearly evident from eq .
( 12 ) , the change of the scattering length depends on the `` starting value '' @xmath72 to which the effect is added . in our calculations ,
cib effects are generated step by step , which implies that the starting value @xmath72 is different for each cib effect .
this distorts the relative size of different cib contributions to the scattering length difference . to make the relative comparison meaningful
, we have rescaled our results for @xmath18 according to a prescription given by ericson and miller @xcite , which goes as follows .
assume the `` starting value '' for the scattering length is @xmath72 and a certain cib effect brings it up to @xmath73 .
then , the resulting scattering length difference @xmath80 is rescaled by @xmath81 with @xmath82 fm and @xmath83 fm . this will make @xmath84 independent of the choice for @xmath72 .
the numbers given in table iii and iv for @xmath18 are all rescaled according to eq .
( 13 ) .
we obtain a total @xmath18 of 4.65 fm which is about 80% of the empirical value of 5.7 fm [ eq .
( 4 ) ] . for @xmath85
we find a total of 0.115 fm from all effects , consistent with the empirical value [ eq .
( 5 ) ] . even though our total result for @xmath18 is very similar to the earlier calculation of ref .
@xcite , there are some differences in the details . for the total effect from @xmath1-exchange we obtain in the present calculations @xmath86 fm while in ref .
@xcite 0.85 fm was reported .
this is due to differences in the interpretation of the scaling mass that occurs in the @xmath87 lagrangian , eq .
( 8) . while in the present calculations we always use @xmath36
[ see discussion below eq .
( 8) ] , in the earlier calculations of ref .
@xcite @xmath38 was used for @xmath39 coupling and @xmath36 for @xmath50 coupling .
the latter convention introduces a strong charge - dependence of the effective @xmath87 coupling strength , which enhances the cib effects from all diagrams involving @xmath27-isobars . in principal
, there is discretion in how to deal with the scaling mass in eq .
however , in the present calculations , we decided to follow the established convention @xcite . as a result ,
the effect from @xmath1-exchange is smaller than in the earlier calculation of ref .
@xcite .
there is also a small difference in the @xmath18 contribution from non - iterative @xmath68 and @xmath69 exchanges for which we obtain 1.4 fm while ref @xcite reported 1.2 fm .
this discrepancy is due to the fact that in ref .
@xcite a preliminary version of the bonn full model @xcite was used , while here we applied the final version @xcite in which the strength of the @xmath68 contribution is slightly larger , which explains the difference .
we note that the cib effect depends on the @xmath88 coupling constant . in the present calculations ,
we follow the bonn model @xcite : we assume charge - independence of the coupling constant and use @xmath89 @xcite . in recent years , there has been some controversy about the precise value of the @xmath88 coupling constant . unfortunately
, the problem is far from being settled .
based upon @xmath0 phase shift analysis , the nijmegen group @xcite advocates the ` small ' charge - independent value @xmath90 , while a very recent determination by the uppsala group @xcite based upon high precision @xmath8 charge - exchange data at 162 mev resulted in the ` large ' value @xmath91 .
other recent determinations are in - between the two extremes : the vpi group @xcite quotes @xmath92 from @xmath93 and @xmath0 analysis with no evidence for charge - dependence .
bugg and machleidt @xcite obtain @xmath94 and @xmath95 from the analysis of @xmath0 elastic data between 210 and 800 mev .
because of this large uncertainty in the @xmath88 coupling constant , it might be of interest to know what the cib effects are like when a value is used that deviates substantially from our choice . for that reason
, we have repeated our cib calculations with the smaller values @xmath96 and 13.6 .
it turns out that the total cib effect on phase shifts ( last column of table i ) as well as the effect on the effective range parameters ( table iii ) scales linearly with the @xmath88 coupling constant , to a good approximation . to be precise : multiplying the total phase shift differences in table i or the effective range changes in table iii with 13.6/14.4 reproduces within @xmath97% the exact results from a cib calculation that employes @xmath98 . as last item in our study
, we have also investigated the effect of rho - mass splitting on the @xmath2 effective range parameters .
unfortunately , the evidence for rho - mass splitting is very uncertain , with the particle data group @xcite reporting @xmath99 mev .
consistent with this , we assumed in our exploratory study @xmath100 mev and @xmath101 mev , i. e. , a splitting of 1 mev . with this , we find @xmath102 fm from one - rho - exchange , and @xmath103 fm from the non - iterative @xmath21 diagrams with @xmath0 intermediate states .
thus , individual effects are small and , in addition , there are substantial cancellations between the two classes of diagrams that contribute .
the net result is a vanishing effect .
thus , even if the rho - mass splitting will be better determined in the future and may turn out to be larger than our assumption , it will never be a great source of cib .
based upon the bonn meson - exchange model for the @xmath0 interaction , we have calulated the cib effects due to pion - mass splitting on the singlet effective range parameters and on the phase shifts of @xmath0 scattering for partial waves of total angular momentum @xmath104 and laboratory energies below 300 mev .
this investigation complements our recent paper on charge - asymmetry of the @xmath0 interaction @xcite .
the largest phase shift differences occur in the @xmath2 state where they are most noticable at low energy ;
e. g. , at 1 mev , the difference is 4.36@xmath105 , indicating that the _ np _ nuclear force is more attractive than the _ pp _ nuclear force .
the @xmath2 phase shift difference decreases with increasing energy and is about 0.6@xmath105 at 300 mev .
the major part of the phase shift difference comes from ope .
cib contributions from two - meson exchange diagrams can be large , but there are typically cancelations between the effects from different classes of diagrams of the two - meson type .
the cib effect on the phase shifts of @xmath106 and higher partial waves is generally small .
the most significant difference is found in the @xmath53 state around 50 mev where the difference is almost 1 degree . in @xmath106-waves ,
the difference is roughly constant above 25 mev : it is 0.95 - 0.65@xmath105 in @xmath53 , about 0.35@xmath105 in @xmath52 , and around 0.2@xmath105 in @xmath5 . in all other partial waves ,
it is in the order of 0.1@xmath105 or less .
again , the main effect comes from ope , however , in @xmath53 , @xmath52 , and @xmath65 at 300 mev , the effect from the @xmath1 model is in the range of 20 - 50% of the one from ope .
the fact that the magnitudes of the phase shift differences in all partial waves , except @xmath2 , are small , makes it difficult to verify experimentally the charge dependent effects in @xmath106 and higher partial waves .
however , since the phase shifts in these states themselves are small , the relative magnitudes of the phase shift differences are not negligible and could have a noticable effect on some sensitive observables such as the analyzing power ( @xmath107 ) in nucleon - deuteron ( @xmath108 ) scattering @xcite since this reaction blows up effects from triplet @xmath106 waves @xcite .
our microscopic predictions are , however , substantially smaller than what is needed to solve the @xmath108 @xmath107 puzzle @xcite .
as mentioned in the introduction , another cib contribution to the nuclear force is irreducible pion - photon ( @xmath109 ) exchange .
traditionally , it was believed that this contribution would take care of the remaining 20% of @xmath18 @xcite . however
, a recently derived @xmath109 potential based upon chiral perturbation theory @xcite _ decreases _ @xmath18 by about 0.6 fm , making the discrepancy even larger .
thus , we are faced with the fact that about 20 - 30% of the charge - dependence of the singlet scattering length is not explained . in recent years
, nuclear physicists have become increasingly concerned with chiral symmetry which is an approximate symmetry of qcd in the light - quark sector . in the light of these new views , the @xmath0 interaction should have a clear relationship with chiral symmetry .
the bonn model that our investigation in based upon is , by construction , not a consistently chiral model .
chiral models for the @xmath0 interaction and , in particular , chiral models for the @xmath1 exchange have recently been constructed by various groups @xcite . however , most of these models are applicable only for the peripheral partial waves of @xmath0 scattering and not for @xmath110 , @xmath106 , or @xmath64 waves ; and if there are predictions for lower partial waves , they are only of qualitative nature . the cib effects in @xmath110 and @xmath106 waves and , particularly , for the singlet scattering length are very subtle and , therefore , require a quantitative model .
thus , current chiral models for the @xmath1 exchange are not ( yet ) suitable for reliable calculations of cib .
one may then raise an interesting question : what has to be changed in the bonn model to make it chiral ?
this question can be answered precisely .
the diagrams in figs . 2 ( a ) and ( b ) of ref .
@xcite have to be added to the bonn model ; that is essentially all .
these diagrams include the weinberg - tomozawa @xmath111 vertex which is a characteristic ingredient of any nonlinear realization of chiral symmetry .
however , it has been found independently by different groups @xcite that the @xmath1 exchange diagrams which include the weinberg - tomozawa vertex make a very small , essentially negligible , contribution to the @xmath0 interaction .
one may then expect that the cib caused by these diagrams is also very small @xcite .
thus , there are reasons to believe that the results of this study may be of broader relevance than what the ( formally ) non - chiral character of our model suggests . of course
, the final and reliable answer of the question under consideration can only come from a ` perfect ' and quantitative chiral model for the @xmath0 interaction that is applicable also in @xmath110 waves and for the calculation of scattering lengths . in view of the problems raised concerning scattering length calculations with chiral models @xcite and in view of the continuing general controversy concerning cutoff _ versus _ dimensional regularization , it will take many years until a reliable calculation of this kind can be done .
thus , for the time being , it may be comforting to have at least our present results .
99 g. a. miller , b. m. k. nefkens , and i. slaus , phys .
reports * 194 * , 1 ( 1990 ) .
g. a. miller and w. h. t. van oers , in _ symmetries and fundamental interactions in nuclei _ , w. c. haxton and e. m. henley , eds .
( world scientific , singapore , 1995 ) p. 127 . c. y. cheung and r. machleidt , phys .
c * 34 * 1181 ( 1986 ) .
r. machleidt , in _ quarks and nuclear structure _ ,
3rd klaus erkelenz symposium , bad honnef , 1983 , lecture notes in physics , vol . 197 , edited by k. bleuler ( springer , heidelberg , 1984 ) , p. 352
t. e. o. ericson and g. a. miller , phys .
* 132b * , 32 ( 1983 ) . m. h. partovi and e. l. lomon , phys .
d * 2 * , 1999 ( 1970 ) .
g. q. li and r. machleidt , _ charge - asymmetry of the nucleon - nucleon interaction _ , nucl - th/9804023 ; phys .
c , in press .
r. machleidt , k. holinde , and ch .
elster , phys .
reports * 149 * , 1 ( 1987 ) .
r. machleidt , adv .
nucl . phys .
* 19 * , 189 ( 1989 ) .
r. machleidt and g. q. li , phys .
reports * 242 * , 5 ( 1994 ) .
particle data group , review of particle properties , phys .
d * 45 * ( 1992 ) .
o. dumbrajs _
_ , nucl . phys . *
b216 * , 277 ( 1983 ) .
the relationship between @xmath112 and @xmath113 [ the latter being used in the lagrangian eq .
( 7 ) ] is @xmath114 following the bonn model @xcite , we also assume a fixed relationship between @xmath115 [ used in the lagrangian eq .
( 8) ] and @xmath113 , namely , @xmath116 v. stoks , r. timmermans , and j. j. de swart , phys .
c * 47 * , 512 ( 1993 ) .
c * 57 * , 1077 ( 1998 ) .
r. a. arndt , r. l. workman , and m. m. pavan , phys .
c * 49 * , 2729 ( 1994 ) ; r. a. arndt , i. i. strakovsky , and r. l. workman , _ ibid .
_ , * 52 * , 2246 ( 1995 ) .
d. v. bugg and r. machleidt , phys .
c * 52 * , 1203 ( 1995 ) .
h. witala and w. glckle , nucl . phys . *
a528 * , 48 ( 1991 ) ; w. glckle , h. witala , d. hber , h. kamada , and j. golak , phys .
reports * 274 * , 107 ( 1996 ) .
i. slaus , r. machleidt , w. tornow , w. glckle , and w. witala , comments nucl .
phys . * 20 * , 85 ( 1991 ) .
d. hueber and j. l. friar , _
the @xmath107 puzzle and the nuclear force _ , nucl - th/9803038 .
u. van kolck , m. c. m. rentmeester , j. l. friar , t. goldman , and j. j. de swart , phys .
lett . * 80 * , 4386 ( 1998 ) . c. ordonez , l. ray , and u. van kolck , phys .
c * 53 * , 2086 ( 1996 ) .
m. r. robilotta and c. a. da rocha , nucl .
a615 * , 391 ( 1997 ) .
n. kaiser , r. brockmann , and w. weise , nucl .
* a625 * , 758 ( 1997 ) .
u. van kolck , privat communication .
k. a. scaldeferri , d. r. philipps , c. w. kao , and t. d. cohen , phys .
c * 56 * , 679 ( 1997 ) .
d. b. kaplan , m. b. savage , and m. b. wise , nucl .
* b478 * , 629 ( 1996 ) .
rrrrrr t@xmath117 ( mev ) & ope & @xmath1 & @xmath21 & @xmath67 & total + + + 1 & 3.051 & 0.319 & -0.361 & 1.355 & 4.364 + 5 & 1.767 & 0.154 & -0.193 & 0.719 & 2.446 + 10 & 1.364 & 0.103 & -0.147 & 0.545 & 1.864 + 25 & 0.944 & 0.047 & -0.107 & 0.391 & 1.275 + 50 & 0.712 & 0.009 & -0.090 & 0.318 & 0.950 + 100 & 0.563 & -0.031 & -0.083 & 0.276 & 0.725 + 150 & 0.519 & -0.061 & -0.086 & 0.267 & 0.638 + 200 & 0.509 & -0.091 & -0.093 & 0.269 & 0.595 + 300 & 0.544 & -0.163 & -0.121 & 0.300 & 0.559 + + + 1 & -0.032 & 0.000 & 0.000 & 0.000 & -0.032 + 5 & -0.246 & -0.003 & 0.000 & -0.001 & -0.250 + 10 & -0.482 & -0.009 & 0.001 & -0.002 & -0.492 + 25 & -0.827 & -0.029 & 0.002 & -0.005 & -0.858 + 50 & -0.902 & -0.053 & 0.005 & -0.009 & -0.960 + 100 & -0.786 & -0.078 & 0.008 & -0.012 & -0.869 + 150 & -0.685 & -0.091 & 0.009 & -0.011 & -0.778 + 200 & -0.618 & -0.101 & 0.011 & -0.009 & -0.717 + 300 & -0.540 & -0.118 & 0.012 & -0.002 & -0.648 + + + 1 & 0.017 & 0.000 & 0.000 & 0.000 & 0.017 + 5 & 0.116 & 0.001 & 0.000 & 0.000 & 0.117 + 10 & 0.203 & 0.002 & 0.000 & 0.001 & 0.206 + 25 & 0.313 & 0.007 & -0.001 & 0.003 & 0.322 + 50 & 0.346 & 0.016 & -0.004 & 0.008 & 0.366 + 100 & 0.323 & 0.032 & -0.009 & 0.018 & 0.364 + 150 & 0.289 & 0.046 & -0.014 & 0.028 & 0.349 + 200 & 0.259 & 0.058 & -0.020 & 0.037 & 0.335 + 300 & 0.213 & 0.083 & -0.032 & 0.057 & 0.321 + + + 5 & -0.009 & 0.000 & 0.000 & 0.000 & -0.009 + 10 & -0.025 & 0.000 & 0.000 & 0.000 & -0.025 + 25 & -0.052 & 0.001 & 0.000 & 0.000 & -0.050 + 50 & -0.045 & 0.005 & 0.000 & 0.000 & -0.040 + 100 & 0.003 & 0.015 & -0.001 & 0.001 & 0.018 + 150 & 0.044 & 0.025 & -0.001 & 0.001 & 0.069 + 200 & 0.071 & 0.034 & -0.002 & 0.002 & 0.106 + 300 & 0.100 & 0.047 & -0.002 & 0.002 & 0.147 + + + 5 & -0.010 & 0.000 & 0.000 & 0.000 & -0.011 + 10 & -0.030 & 0.000 & 0.000 & -0.001 & -0.032 + 25 & -0.096 & -0.001 & -0.001 & -0.003 & -0.101 + 50 & -0.172 & -0.003 & -0.002 & -0.007 & -0.184 + 100 & -0.224 & -0.007 & -0.005 & -0.014 & -0.250 + 150 & -0.224 & -0.007 & -0.007 & -0.019 & -0.258 + 200 & -0.210 & -0.007 & -0.008 & -0.022 & -0.248 + 300 & -0.183 & -0.004 & -0.010 & -0.022 & -0.219 + + + 10 & -0.004 & 0.000 & 0.000 & 0.000 & -0.004 + 25 & -0.020 & 0.000 & 0.000 & 0.000 & -0.020 + 50 & -0.045 & 0.000 & 0.000 & 0.000 & -0.045 + 100 & -0.070 & -0.001 & 0.000 & 0.000 & -0.071 + 150 & -0.084 & -0.001 & 0.000 & 0.000 & -0.084 + 200 & -0.093 & -0.001 & 0.000 & 0.000 & -0.093 + 300 & -0.102 & -0.001 & -0.001 & 0.001 & -0.102 + + + 5 & 0.012 & 0.000 & 0.000 & 0.000 & 0.012 + 10 & 0.036 & 0.000 & 0.000 & 0.000 & 0.036 + 25 & 0.091 & 0.001 & 0.000 & 0.000 & 0.092 + 50 & 0.119 & 0.002 & 0.000 & 0.001 & 0.121 + 100 & 0.095 & 0.007 & -0.001 & 0.001 & 0.102 + 150 & 0.057 & 0.011 & -0.002 & 0.002 & 0.068 + 200 & 0.025 & 0.014 & -0.003 & 0.003 & 0.038 + 300 & -0.018 & 0.018 & -0.006 & 0.003 & -0.003 + + + 10 & 0.009 & 0.000 & 0.000 & 0.000 & 0.009 + 25 & 0.044 & 0.000 & 0.000 & 0.000 & 0.044 + 50 & 0.093 & 0.000 & 0.000 & 0.000 & 0.093 + 100 & 0.141 & 0.001 & 0.000 & 0.000 & 0.142 + 150 & 0.159 & 0.003 & 0.000 & 0.000 & 0.163 + 200 & 0.166 & 0.005 & 0.000 & 0.001 & 0.171 + 300 & 0.165 & 0.010 & -0.001 & 0.002 & 0.177 + + + 25 & -0.009 & 0.000 & 0.000 & 0.000 & -0.009 + 50 & -0.023 & 0.000 & 0.000 & 0.000 & -0.023 + 100 & -0.032 & 0.001 & 0.000 & 0.000 & -0.032 + 150 & -0.029 & 0.002 & 0.000 & 0.000 & -0.027 + 200 & -0.022 & 0.003 & 0.000 & 0.000 & -0.019 + 300 & -0.005 & 0.006 & 0.000 & 0.000 & 0.001 + + + 25 & -0.004 & 0.000 & 0.000 & 0.000 & -0.004 + 50 & -0.015 & 0.000 & 0.000 & 0.000 & -0.015 + 100 & -0.037 & 0.000 & 0.000 & 0.000 & -0.037 + 150 & -0.053 & 0.000 & 0.000 & 0.000 & -0.053 + 200 & -0.064 & 0.000 & 0.000 & -0.001 & -0.065 + 300 & -0.078 & -0.001 & 0.000 & -0.001 & -0.080 + + + 50 & -0.007 & 0.000 & 0.000 & 0.000 & -0.007 + 100 & -0.019 & 0.000 & 0.000 & 0.000 & -0.019 + 150 & -0.028 & 0.000 & 0.000 & 0.000 & -0.029 + 200 & -0.035 & 0.000 & 0.000 & 0.000 & -0.036 + 300 & -0.044 & -0.001 & 0.000 & 0.000 & -0.045 + + + 25 & 0.012 & 0.000 & 0.000 & 0.000 & 0.012 + 50 & 0.033 & 0.000 & 0.000 & 0.000 & 0.033 + 100 & 0.058 & 0.000 & 0.000 & 0.000 & 0.059 + 150 & 0.067 & 0.001 & 0.000 & 0.000 & 0.068 + 200 & 0.068 & 0.001 & 0.000 & 0.000 & 0.070 + 300 & 0.062 & 0.002 & 0.000 & 0.000 & 0.065 + rrrrrrrr t@xmath117 ( mev)&@xmath54-@xmath110 & @xmath54-@xmath118&@xmath57-@xmath119 & @xmath57-@xmath118 & @xmath61-@xmath119 & @xmath61-@xmath118 & total @xmath1 + + + 1 & -0.182 & 0.319 & 0.692 & -0.410 & -0.244 & 0.143 & 0.319 + 5 & -0.098 & 0.160 & 0.363 & -0.217 & -0.130 & 0.076 & 0.154 + 10 & -0.075 & 0.113 & 0.271 & -0.165 & -0.099 & 0.058 & 0.103 + 25 & -0.056 & 0.065 & 0.188 & -0.119 & -0.072 & 0.042 & 0.047 + 50 & -0.047 & 0.037 & 0.146 & -0.099 & -0.061 & 0.034 & 0.009 + 100 & -0.045 & 0.012 & 0.118 & -0.091 & -0.056 & 0.030 & -0.031 + 150 & -0.047 & -0.003 & 0.109 & -0.093 & -0.058 & 0.030 & -0.061 + 200 & -0.052 & -0.015 & 0.107 & -0.100 & -0.062 & 0.031 & -0.091 + 300 & -0.067 & -0.041 & 0.112 & -0.126 & -0.078 & 0.037 & -0.163 + + + 10 & -0.003 & -0.003 & 0.001 & -0.003 & 0.000 & 0.000 & -0.009 + 25 & -0.010 & -0.010 & 0.002 & -0.009 & -0.001 & 0.000 & -0.029 + 50 & -0.018 & -0.020 & 0.002 & -0.017 & -0.002 & 0.000 & -0.053 + 100 & -0.025 & -0.030 & 0.003 & -0.024 & -0.002 & 0.001 & -0.078 + 150 & -0.027 & -0.036 & 0.002 & -0.028 & -0.003 & 0.001 & -0.091 + 200 & -0.029 & -0.041 & 0.002 & -0.030 & -0.003 & 0.001 & -0.101 + 300 & -0.031 & -0.050 & 0.000 & -0.035 & -0.004 & 0.001 & -0.118 + + + 10 & 0.000 & 0.002 & 0.001 & -0.001 & 0.000 & 0.000 & 0.002 + 25 & -0.001 & 0.006 & 0.004 & -0.002 & 0.000 & 0.000 & 0.007 + 50 & -0.002 & 0.012 & 0.010 & -0.003 & -0.001 & 0.001 & 0.016 + 100 & -0.003 & 0.021 & 0.021 & -0.006 & -0.001 & 0.001 & 0.032 + 150 & -0.004 & 0.028 & 0.030 & -0.008 & -0.002 & 0.002 & 0.046 + 200 & -0.004 & 0.034 & 0.038 & -0.009 & -0.002 & 0.002 & 0.058 + 300 & -0.004 & 0.043 & 0.054 & -0.010 & -0.003 & 0.003 & 0.083 + + + 50 & 0.000 & 0.003 & 0.002 & 0.000 & 0.000 & 0.000 & 0.005 + 100 & 0.000 & 0.009 & 0.007 & -0.001 & 0.000 & 0.000 & 0.015 + 150 & 0.000 & 0.014 & 0.013 & -0.002 & -0.001 & 0.001 & 0.025 + 200 & 0.000 & 0.017 & 0.019 & -0.003 & -0.001 & 0.001 & 0.034 + 300 & 0.000 & 0.020 & 0.031 & -0.004 & -0.002 & 0.002 & 0.047 + + + 10 & -0.001 & 0.001 & 0.001 & -0.001 & 0.000 & 0.000 & 0.000 + 25 & -0.003 & 0.002 & 0.005 & -0.004 & -0.001 & 0.001 & -0.001 + 50 & -0.006 & 0.004 & 0.011 & -0.010 & -0.004 & 0.002 & -0.003 + 100 & -0.009 & 0.004 & 0.020 & -0.018 & -0.008 & 0.004 & -0.007 + 150 & -0.010 & 0.004 & 0.027 & -0.022 & -0.011 & 0.005 & -0.007 + 200 & -0.010 & 0.003 & 0.031 & -0.024 & -0.014 & 0.007 & -0.007 + 300 & -0.009 & 0.003 & 0.037 & -0.024 & -0.018 & 0.008 & -0.004 + .cib contributions to the @xmath2 scattering length , @xmath18 , and effective range , @xmath85 , from various components of the @xmath0 interaction as explained in the text .
[ cols=">,>,>,>,>,>,>",options="header " , ]
|
based upon the bonn meson - exchange - model for the nucleon - nucleon ( @xmath0 ) interaction , we calculate the charge - independence breaking ( cib ) of the @xmath0 interaction due to pion - mass splitting . besides the one - pion - exchange ( ope ) , we take into account the @xmath1-exchange model and contributions from three and four irreducible pion exchanges .
we calculate the cib differences in the @xmath2 effective range parameters as well as phase shift differences for partial waves up to total angular momentum @xmath3 and laboratory energies below 300 mev .
we find that the cib effect from ope dominates in all partial waves .
however , the cib effects from the @xmath1 model are noticable up to d - waves and amount to about 40% of the ope cib - contribution in some partial waves , at 300 mev .
the effects from 3@xmath4 and 4@xmath4 contributions are negligible except in @xmath2 and @xmath5 .
|
ices are important constituents in the collisional formation of comets and planetesimals in protoplanetary disks .
as various species of ice appear in different distances to the central star due to their individual sublimation pressures , there are various icelines within the disk .
planet formation in general but especially collision outcomes are tied to these icelines and the physics of the prevailing kind of ice . @xcite . in earlier experiments we showed that it is necessary to distinguish the different ices while studying their growth potential @xcite .
we suggested that collisional growth should be most efficient between the water and the co@xmath0 iceline or between 2.0 to 9.3 au according to the minimum mass solar nebula model .
beyond 9.3 au non - polar co@xmath0 dominates .
collision experiments showed that co@xmath0 collisions are comparable to silicate collisions .
an increased sticking of water ice was proposed but not verified in experiments .
h@xmath0o is often considered as a driving mechanism for planetesimal growth due to a high sticking efficiency compared to other materials @xcite .
while this is plausible as water does have a high electrical dipole moment and as individual experiments with water ice point to this , a direct experimental comparison with other ices under similar conditions is missing .
therefore , we compare collisions of pure co@xmath0 , pure h@xmath0o and 1:1 mixtures by mass here .
the experimental setup is equal to the former experiment from @xcite .
details can be found there . in short ,
the experiment consists of a vacuum chamber flooded with different gas samples . in the former experiment we used co@xmath0-gas . in this work
we additionally take a mixture between co@xmath0 and h@xmath0o gases and h@xmath0o vapor . a scheme of the experimental setup
is shown in fig .
[ kryostat ] .
the chamber is cooled by liquid nitrogen to a temperature of 80 k in the collision section . within several minutes
a 2 mm thick ice layer is deposited on the walls of the chamber .
the chamber is sealed and evacuated to an ambient pressure of @xmath21 mbar . by means of a gearwheel driven by an electrical motor ice grains
are beveled off from the cryostate .
this produces @xmath290 @xmath3 m ice particles .
the chemical composition of the produced aggregates depends on the gas ( mixture ) which was used to float the chamber .
the particles collide with an ice - layer of the same composition which is deposited on a copper plate connected to the cryostate .
collision velocities reach up to 1 m / s in this work .
collisions are imaged with a camera at 1250 frames per second with a spatial resolution of 10 @xmath3 m .
experimental setup : a vacuum chamber is flooded with a gas ( mixture ) which is deposited on a cryostate cooled with liquid nitrogen . then the chamber is evacuated to an ambient pressure of 0.1 - 1 mbar and particles are beveled off by using a rotating gearwheel .
the particles collide with an ice layer on a copper plate connected to the cryostate . depending on the impact velocity , sticking or bouncing can be observed . in earlier work
we also observed fragmentation for co@xmath0 @xcite . ]
the colliding ice particles are generated by scraping them off from the cooled cryostate .
the structure and the chemical composition of these aggregates depends on the initial atmosphere within the vacuum chamber . for pure co@xmath0
aggregates the chamber is just continuously flooded by a stream of co@xmath0 gas .
the co@xmath0 sublimation point is at about 195 k @xcite .
therefore , we argue that the molecules only freeze out once they hit a cold wall but do not already form ice grains within the gas phase .
the scraping then removes material from a solid ice surface . according to our earlier analysis of fragmenting collisions the co@xmath0 particles are small aggregates . for the h@xmath0o - co@xmath0 aggregates we use the same procedure with a mixture of co@xmath0 gas and h@xmath0o vapor .
the mass ratio between co@xmath0 and h@xmath0o is 1:1 .
this value was determined after the experiment by weighing the mass loss of the ice mixture outside of the chamber , which is caused by the rapid evaporation of co@xmath0 . unlike the co@xmath0 vapour the water supply already consists of @xmath4 m water droplets besides vapour .
the structure of the ice surface will be slightly different then .
we currently can not analyze this in detail but would argue on the state of the ice surface as follows : some of the water droplets will diffuse towards the walls and stick there as water ice grains .
however , gas molecules will reach the surface by diffusion preferentially ( due to their smaller mass compared to the droplets ) and stick there as in the pure co@xmath0 case .
this leads to an intimate mixture of co@xmath0 and water ice molecules with a certain amount of pure water ice grains embedded which can be regarded as homogeneous material .
our collision results are in favor of this view .
as seen below , we see a clear threshold velocity between bouncing and sticking for particles from the mixed sample .
this threshold is distinguished from the pure co@xmath0 threshold by an order of magnitude .
would the particles which are beveled off be inhomogeneous and consist of large fractions of pure water or pure co@xmath0 ice or otherwise locally distinguished compositions at their surface , then an individual collision should depend on the material during contact which should e.g. be water or co@xmath0 in a given collision . in any case we should not see a clear threshold in collision experiments then .
the same argument holds if there is a significant amount of ice grains embedded but if scraping preferentially breaks the mixed matrix in between which is supposedly less stable .
also then only one sort of mixed surface would interact with the same material . in this case
the mixing ratio might not be the mixing ratio of the matrix material measured ( 1:1 ) though , but might be shifted towards the co@xmath0 fraction .
as the co@xmath0 fraction is high in this case the mix will glue together any pure water ice spheres efficiently , though .
they might be porous , but the aggregates can not restructure at the given impact energies and can be considered as individual grains for low velocities . for the h@xmath0o sample we flooded the chamber only with water vapour and droplets .
the same arguments hold that there should be a mixture of water ice droplets embedded in a water ice matrix .
it is likely though that the water ice matrix is very thin and weak . in this case
it might not be appropriate to treat the aggregate particles as single grains as already small impact velocities might be sufficient to restructure the aggregate .
some elasticity in larger aggregates is clearly visible in the high speed observations for the case of pure water ice .
this changes the outcome of collisions ( e.g. @xcite ) .
it will depend on the specific configuration now and the amount of energy that is dissipated by restructuring if an aggregate sticks or bounces .
we should not expect a clear threshold velocity then .
we consider the pure co@xmath0 particles as mixed particles as solid spheres even if they would be aggregates .
all particles are initially modeled as spheres of a size with the same cross sections as the observed grains .
we determined the cross section by optical imaging . this way we get size distrubutions of the particles which can be found in fig .
[ size ] .
size distrubution for the different species of ice particles considered for this study .
the distribution for the co@xmath0 aggregates is taken from @xcite .
the black line marks the average size of 90 @xmath3 m .
a bin size of 20 @xmath3 m is chosen with respect to the spatial resolution of the camera . ]
these distributions are comparable ; they all peak around @xmath250 @xmath3 m and have a medium size of @xmath290 @xmath3 m .
the bin size of 20 @xmath3 m is chosen due to the spatial resolution of the optical system of 10 @xmath5 .
there is an indication in the size distribution of a cut - off at 20 @xmath5 . in any case
, @xcite showed that there is a smallest fragment in a dissipative process so it is not likely for all particles to be covered by a layer of small unseen grains .
smaller particles may be present but should not affect the observed collision outcomes .
the analized grain sizes are a narrow but representative illustration of dust found in protoplanetary disks . here , the sizes reach from sub-@xmath6 , which is a typical size for dust in the ism @xcite , to the cm - regime at which the bouncing barrier sets in @xcite .
the size distribution typically follows a power law @xcite . since the distributions of all analyzed paticles ( fig .
[ size ] ) are comparable we can directly compare the collisional behavior for one mean size of the different species of ice grains depending on the collision velocity keeping in mind though that aggregation might affect the result .
collision outcomes at low speed can be categorized in two types : 1 ) sticking between the grain and the target at low velocities ; 2 ) bouncing from the target at higher impact velocities @xcite .
the threshold between sticking and bouncing is not only important for understanding whether collisional growth of aggregates in protoplanetary disks is possible but also allows to determine fundamental parameters like the surface energy of the particles @xcite . in general
, there are more effects for higher impact velocities like the fragmentation of or mass transfer to the aggregates @xcite which we do not consider in this work . in fig .
[ surface ] we show samples of the largest aggregates for each dataset of different ices ( not analyzed collision - wise ) . for the largest aggregates , the surface structure and porosity can be determined easiest .
observation of these particles suggest a more irregular surface structure and higher porosity for water ice . a quantitative analysis of the surface and the porosity is not possible so far .
as described above , due to the variation in the microstructure between the co@xmath0-aggregates and the co@xmath0-h@xmath0o-/ h@xmath0o - aggregates a comparision of their collisional behavior is given only roughly .
the porous structure of the water ice aggregates suggests that restructuring might be important as indicated above .
the mixed ice samples qualitatively look compact enough that they should only show hit - and - stick or just bouncing slightly above the threshold for sticking .
more quantitatively , the collisional behavior can be described by the coefficient of restitution ( cor ) which we define as the ratio of the absolute value of the particle velocity after @xmath7 and before @xmath8 the collision @xmath9 there are many theroretical models describing the cor for elastic , plastic and viscous spheres @xcite .
other approaches try to describe the cor by heuristic functions , like @xcite did for water ice spheres . for the cor resulting from the observed collisions we use our model deduced in the earlier work on pure co@xmath0 collisions @xcite . @xmath10 with the critical velocity @xmath11 , the sticking velocity @xmath12 and material dependend parameters @xmath13 and @xmath14 . here
, the sticking velocity @xmath12 describes the transition between the sticking and the bouncing regime , where @xmath15 and @xmath16 .
the critical velocity @xmath11 describes the maximum for the cor , where the bouncing is most elastic .
the cor from eq .
models three different effects ; sticking with @xmath17 , elastic bouncing with @xmath18 and plastic deformation with @xmath19 .
this function gives the most appropriate fit for the cor for co@xmath0 particles .
it also fits the cor for the h@xmath0o / co@xmath0 mixture well like shown in fig .
[ cor_mix ] . the coefficient of restitution ( cor ) for the mixture particles with a h@xmath0o - co@xmath0 mass - ratio of 1:1 .
the dashed line describes the cor for pure co@xmath0 particles from @xcite .
the solid line represents the fit function from eq . .
the uncertainties result from uncertainties in the impact velocity and the aggregate size . ]
the fit parameters for both fits are summarized in tab .
[ fitpar ] .
.fit parameters for the cor . [ cols="^,^,^",options="header " , ] the values for co@xmath0 particles are taken from @xcite .
the gravitational force is acting on the aggregates during collisons .
this effect distorts the analysis of the data , because low cor collisions might be classified as sticking events .
nonetheless , in @xcite we show , that this effect becomes significant only for collisions with impact velocities below 0.05 m / s .
this is also the reason for deducing the sticking velocity for co@xmath0 aggregates by a model fit in the previous work .
collisions of co@xmath0-h@xmath00 particles are more inelastic than collisions of co@xmath0 , since the cor in fig .
[ cor_mix ] has smaller values for the mixture .
moreover the sticking velocity is an order of magnitude larger . from the sticking velocity
we can calculate the surface energy for co@xmath0-h@xmath00 particles from @xcite , @xmath20 with the surface energy @xmath21 , the reduced radius @xmath22 , the particle mass density @xmath23 and an elastic constant @xmath24 , where @xmath25 . in this definition @xmath26
is the young s modulus and @xmath27 is the poisson s ratio . taking a mean young s modulus between water ice and co@xmath0 ice of @xmath28 gpa , a mean density of @xmath29 kg / m@xmath30 and a mean poisson ratio of @xmath31 @xcite we get a surface energy for particles with a mean radius of @xmath32 m of @xmath33 we determined the surface energy for pure co@xmath0 ice in our earlier work to @xmath34 @xcite .
compared to this the surface energy of mixed particles this is an order of magnitude higher . for the h@xmath0o aggregates , we do not see a sharp transition from the sticking regime to the bouncing regime which might be due to the preparation and restructuring of the aggregates in collisions .
[ cor_w ] shows the measured data for the coefficient of restitution . the coefficient of restitution ( cor ) for water ice particles .
the dashed lines describe the threshold for co@xmath0 and co@xmath0-h@xmath0o data . ]
we can determine a sticking probability @xmath35 though , defined as the ratio between all sticking events in collisions @xmath36 and all analyzed collison events @xmath37 for certain velocity range .
@xmath38 for impact velocities between 0 - 0.5 m / s the sticking probability is @xmath39 and for impact velocities between 0.5 - 1 m / s @xmath40 . for comparability of pure water ice
, one can use eq . to get a rough estimate of the dependency between the h@xmath0o fraction @xmath41 in the particles and the sticking velocity @xmath42 for the range
@xmath43 $ ] . assuming a simple linear approximation of @xmath44 with the constants @xmath45 we obtain @xmath46 using eq .
we can fit the sticking velocities determined for @xmath47 and @xmath48 and extrapolate this function for an h@xmath0o fraction of 1 .
for the constants we get @xmath49 and @xmath50 m / s .
this procedure leads to a value of @xmath51 m / s ( fig .
[ threshold ] ) . within the range
studied we do see sticking and bouncing of pure water ice at this velocity .
the measured water data is therefore at least not in contradiction to this extrapolation . in their recent work
@xcite give a sticking threshold for 1.5 @xmath3 m sized water ice of 9.6 m / s .
if we scale this value with the dependency from eq .
, @xmath52 , we get a sticking velocity of 0.31 m / s for 90 @xmath3 m pure water ice grains . within the range of unknowns ( porosity , regular - irregular grain relation ) our data
are also consitent with this .
threshold velocities between the sticking and the bouncing regime depending on the h@xmath0o fraction of the ice .
the black dots show measured values and the circle an estimation from eq . .
the black line shows the fit function which changes over to the extrapolation . ]
fig . [ threshold ] clearly shows that threshold velocities for sticking of h@xmath0o / co@xmath0 mixtures vary between the less sticky end member co@xmath0 and the much stickier other end of pure h@xmath0o ice .
the analytical dependence between @xmath21 and @xmath41 chosen or in general the sticking velocity dependence on @xmath41 could be more complex though .
in general , the jkr - theory can give us only an approximation here anyway , because it does not describe nonspherical aggregate shapes and complicated contacts including dipole moments or hydrogen bonds . a more detailed analysis based on thorough variation of the mass ratios ( @xmath41 )
would be needed to quantify this further . however , as outlined above the preparation does not warrant homogeneous mixtures yet but grains might e.g. be core - mantle particles with embedded water ice grains . in this case
the total mass ratios determined by sublimation of the co@xmath0 afterwards could be different from mass ratios of the surfaces which are actually in contact during a collision .
hence , such a study would require a spatially resolved analysis of the composition of the grains on the nm - scale .
this seems feasible with current microscopic techniques but due to the volatile nature of the ices this is not straightforward and is beyond the scope of this paper .
we used the sticking velocity measured and modeled as given and deduced the surface energy based on eq .
[ stickvel ] .
this might lead to a mismatch with given literature data .
values for the surface energy of pure h@xmath0o ice are e.g. 0.37 j / m@xmath53 from @xcite or 0.19 j / m@xmath53 from @xcite
. our value of the mixed sample should be below the water value and the determined value of 2.8 j / m@xmath53 looks to be too large then ( by an order of magnitude ) .
our deduced values for pure co@xmath0 is also overestimating existing values by a factor of a few @xcite . experimental work from @xcite suggested that the model by @xcite might give values for @xmath12 which are an order of magnitude too low .
if this is a theoretical ( model ) / experimental mismatch , our deduced values from eq . could be overestimated by an order of magnitude though they would be self - consistent among themselves .
another uncertainty in the calculation of the surface energy is the structure of the particles .
for once they are irregular and not spheres and likely can be small aggregates . with the optical observation used we can not resolve the particles further and do not know the total mass exactly .
we used bulk densities instead . in total , the value in eq .
should rather be treated as an estimation than an exact result at the moment .
this does not devalue the qualitative dependence of the sticking velocity with water content though .
in this work we measured the threshold velocity between sticking and bouncing for collisions of ice grains with a solid wall of the same material .
we studied three compositions : pure dry ice , pure water ice and a 1 to 1 mixture . as we used the same setup , the results are immediately comparable . in all three cases
the average particle size analyzed was @xmath54 @xmath5 @xmath55 @xmath5 .
collision velocities were up to 1 m / s .
our goal was to show explicitly how in comparison the sticking properties of these ices differ in collisions .
while co@xmath0 particles of the given size only stick at velocities below 0.04 m / s , mixtures of 1:1 mass ratio have a sticking threshold at 0.43 m / s .
for pure h@xmath0o ice we could not find a steep threshold velocity but only probabilities on the order of 60 to 70 % for sticking up to 1 m / s , which is likely caused by a fragile aggregate structure . from the sticking velocity for the particles consisting of the mixture of h@xmath0o / co@xmath0 ice we calculate the surface energy to @xmath56 @xmath57 .
compared to pure co@xmath0 ice this is an order of magnitude higher , though the quantitative derivation of this values should be considered with care . in our earlier paper
we showed that co@xmath0 behaves mostly like silicates @xcite . here , we show that adding water makes particles much stickier .
this supports the idea that there is an inner silicate dominated region and and outer co@xmath0 dominated region in protoplanetary disks where collisional growth might be less efficient than in the region in between the respecitve ice lines where water ice dominates .
this work is supported by the dfg under the grant number wu321/12 - 1 and te890/1 - 1 .
|
collisions of ice particles play an important role in the formation of planetesimals and comets . in recent work we showed , that co@xmath0 ice behaves like silicates in collisions .
the resulting assumption was that it should therefore stick less efficiently than h@xmath0o ice . within this paper a quantification of the latter
is presented .
we used the same experimental setup to study collisions of pure co@xmath0 ice , pure water ice and 50% mixtures by mass between co@xmath0 and water at 80k , 1 mbar and an average particle size of @xmath1 m .
the results show a strong increase of the threshold velocity between sticking and bouncing with increasing water content .
this supports the idea that water ice is favorable for early growth phases of planets in a zone within the h@xmath0o and the co@xmath0 iceline .
|
as discussed earlier in this conference by dave meier , the theoretical framework for the formation of relativistic jets due to the accretion of matter onto a supermasive black hole is well developed .
observations of jet kinematics obtained by multi - epoch high resolution imaging provides tests of theoretical models and can constrain the parameters of the jet formation . however , the interpretation is not straightforward .
the observations do not give instantaneous multi - epoch snapshots of the source structure .
rather , due to relativistic effects , different parts of the source are observed at different times , giving the illusion of superluminal motion , the distortion of angles and misrepresentation of luminosity . in particular , the observed speeds and luminosities of jets are highly biased due to doppler boosting within a narrow cone , and they are not representative of the intrinsic distributions .
finally , we note that apparent speeds derived from multi - epoch vlbi observations refer to the observed motion of recognizable features , typically referred to as `` components , '' which do not necessarily reflect the underlying bulk relativistic flow , but which instead may simply reflect the propagation of forward and reverse shocks within the jet flow .
the observations are further complicated since these features do not remain stable with time , but may break up into sub components , while other features may merge into a single complex feature e.g , @xcite .
prior to the completion of the vlba , observations of jet kinematics were based on infrequent multi - epoch vlbi observations using ad - hoc vlbi arrays generally with too few antennas to obtain the high - sensitivity and high dynamic range imaging needed to unambiguously trace jet motions .
a number of vlba programs have now been underway for up to a decade or more at wavelengths between 7 mm and 13 cm and are providing new insight into jet kinematics .
this paper reports on global source kinematics primarily obtained from 2 cm vlba observations , but we also discuss the multi - wavelength kinematics of a few particularly interesting sources .
@xcite have combined their 6 cm multi - epoch observations of compact radio sources with pre - vlba observations to study the kinematics of 266 radio sources .
typically 2 to 4 epochs were observed for each source .
@xcite have used the radio reference frame image database ( rrfid ) to determine source kinematics for 87 sources at 4 cm over the period 1994 to 1998 .
this is part of a database originally obtained for precise astrometry and to study earth motions and includes multi - epoch observations at both 4 and 13 cm for up to 500 sources .
@xcite and @xcite have reported 2 cm vlba studies for 110 agn jets covering the period 1994 to 2001 .
ros et al . and lister et al .
( in preparation ) extend the observations to 2003 and 2007 respectively .
since 2002 , the mojave program ( monitoring of jets in active galactic nuclei ) has further extended this program to include the kinematics of a complete sample of the the brightest 133 sources in the northern sky @xcite as well as 59 other sources of particular interest . since 1994 , these programs have observed as many as 30 epochs for some sources , permitting for the first time clear evidence for non radial motions , bends , accelerations and decelerations .
the mojave program also includes linear @xcite and circular @xcite polarization measurements which can be used to investigate three - dimensional magnetic structure .
the mojave program is a large vlba program and images of new and past 2 cm vlba observations are made available on the mojave web site within a few weeks following the correlation of the data . in a series of papers
, the boston university group has reported on 0.7 and 1.3 cm vlba observations @xcite .
these observations were made more frequently , typically monthly or bi - monthly , and consequently include a much smaller number of sources than observed by the longer wavelength programs .
monthly observations of 29 sources at 7 mm are continuing with special attention to radio counterparts of egret gamma - ray sources .
recently tanami , ( tracking active galactic nuclei with australian - south - african milliarcsecond interferometry ) a new vlbi program to monitor extragalactic jets in a sample of sources below -30 degrees declination began to observe in november 2007 at 8.4ghz and 22ghz with the australian south - african long baseline array @xcite .
it is planned to expand this sample to include glast detected sources .
due to relativistic beaming the observed brightness temperature is enhanced by the doppler factor ( @xmath0 ) .
the maximum brightness temperature or lower limit that can be determined by vlbi depends only on flux density , measurement accuracy and baseline length , and is independent of the wavelength used .
see kovalev ( these proceedings ) . for a source of @xmath1 1 jy and earth baselines ,
this limit is @xmath2 k. using the vlba at 2 cm , @xcite have observed brightness temperatures up to @xmath3 k , while @xcite have reported values of @xmath4 up to @xmath5 k using halca at 6 cm .
this suggests doppler factors up to a few hundred if @xmath6 is limited by inverse compton cooling to values @xmath7 k. however , the highest observed lorentz factors are only 35 to 40 @xcite , much less than required to explain the highest observed brightness temperatures , unless these high observed values of @xmath8 are only transient .
@xcite have compared the distribution of observed brightness temperatures with relativistic beaming models and find that in their medium to low brightness state , most sources have @xmath9 k or close to an equipartition value . however , when they are in a flaring state , @xmath10 k , which is close to the ic limit .
this correponds to an excess of particle over magnetic density of as much as @xmath11 or more .
@xcite and @xcite have used the characteristic variability time - scales to estimate intrinsic source size and brightness temperature , and find that the distribution of observed speeds and brightness temperature are consistent with @xmath6 being close to the equilibrium value .
@xcite have shown that the distribution of observed speed and luminosity is consistent with relativistic beaming models having a maximum lorentz factor @xmath12 , and intrinsic luminosity @xmath13 whz@xmath14 .
sources with low apparent radio luminosity and slow speeds are unlikely to be powerful highly relativistic jets oriented near the plane of the sky ; nor can they be due to low luminosity high speed jets pointed close to the line of sight as the probability of these configurations is too low . with the extensive data base now available ,
many jets show clear curvature .
a major question is to find out if the jet flow follows a preset curved trajectory or is ballistic .
ballistic motion can produce a curved jet if it originates from a rotating nozzle , as might be induced if the agn is part of a binary system .
however , there is no clear model which explains all observations . in some sources like @xmath15 , successive components appear to follow the same curved trajectory as previously ejected features @xcite
however , in other sources , such as 1308 + 328 , @xcite has shown that the observed pattern is consistent with that expected from a precessing nozzle having a @xmath16 year period .
although gamma - ray emission is associated with flat spectrum radio blazars , the limited sensitivity and poor time resolution of the existing egret observations has precluded a clear understanding of the relation between the radio and gamma ray emission .
@xcite found that gamma
ray blazars have more compact structure than non gamma
ray blazars , while @xcite and @xcite reported somewhat faster observed speeds for gamma ray detected sources .
@xcite suggest that gamma
ray flares occur near the time of ejection of new superluminal components .
m87 is one of the closest known relativistic jets and is of special interest since 1 mas corresponds to a linear scale of only 0.08 pc and an angular velocity of 1 mas / yr corresponds to a linear velocity of 0.25 c. as early as 1964 , @xcite argued that the jet in m87 was intrinsically two - sided but appeared one - sided due to differential doppler boosting of the approaching and receding relativistic jets .
this was the first suggestion that bulk relativistic motion might play an important role in understanding extragalactic radio sources . however ,
curiously @xcite find no evidence at 2 cm over a period of 12 years for any motions in any of seven apparent features greater than a few percent @xmath17 within about 1.5 pc of the jet base .
however , @xcite report superluminal motion in the optical feature known as hst-1 located 0.86 arcsec ( 60 pc ) downstream , while @xcite report @xmath18 in the 20 cm radio feature associated with hst-1 . also , @xcite have reported strong and variable tev emission from m87 , and @xcite argue that the tev emission may originate in hst-1 and not at the base of the jet . closer in to the base
, 7 mm observations by @xcite with submilliarcsecond resolution suggests speeds which are a significant fraction of @xmath17 .
the high linear resolution 2 cm image of @xcite shows the beginning of the counter - jet .
kovelev et al .
suggested that the apparent limb brightening of the main jet is due to a two component spine - sheath configuration as expected from theoretical models @xcite . in the context of these models ,
the 2 cm observations see only the slowly moving outer sheath , while the faster moving inner spine is responsible for the observed high energy emission . however , @xcite speculate that the tev emission might be produced in the slowly moving outer layer surrounding a fast misaligned spine .
3c 279 is one of the strongest radio blazers and its jet kinematics have been studied for more than 30 years .
@xcite have reported an abrupt change in the trajectory of the brightest jet feature with an apparent change in direction of @xmath19 deg and corresponding velocity change from 8@xmath17 to 13@xmath17 .
@xcite concluded that the intrinsic bend was only about 1 degree , but occured more than 1 kpc away from the base of the jet , and that the focusing and collimation of relativistic jets can occur well downstream .
3c 111 is one of only a few known classical double lobed radio galaxies with observed superluminal motion in the core . @xcite
have studied the evolution of the one sided radio jet in 3c 111 , over a period of about 10 years during a major flare and jet - plasma ejection , and found a complex pattern of separating components possibly indicative of forward- and backward directed shocks as well as trailing components formed in the wake of the primary perturbation .
such effects may be interpreted as a result of coupling to kelvin - helmholtz instability pinching modes from the interaction of the jet with the external medium .
they complicate the interpretation of multi - epoch vlbi jet observations and provide a challenge for the descriptive analysis of jet structure as well as for numerical jet models . "
although the parent population of agn jets is mostly only mildly relativistic , in flux density - limited samples , doppler boosting causes the observed distribution to be highly biased toward the fastest jets with observed lorentz factors up to 35 or 40 which are aligned close to the line - of - sight . however , subluminal and stationary features are also observed . although each jet appears to have a characteristic velocity which may reflect the bulk flow , quantitative analysis may be complicated by differences between the bulk plasma flow and observed pattern motions , by the apparent splitting and merging of individual features , by finite opening angles , and by transverse velocity gradients . on average quasars have larger lorentz factors ( @xmath20 ) than low luminosity agn , and egret detected blazars may have somewhat larger lorentz factors .
the observed distribution of lorentz factor and luminosity is consistent with doppler boosting models , but there appear to be intrinsic differences as well as projection effects
. both ballistic motion characteristic of a rotating nozzle and curved trajectories with accelerations and decelerations are observed , and there is some evidence that in
situ acceleration along the jet may be important .
the ejection of new features may be associated with radio or gamma ray outbursts , but the evidence for this is inconclusive .
brightness temperatures up to @xmath21 k are observed directly and are also inferred from the time scale of flux density variability . over the course of the next few years
, the bandwidth of the vlba will be increased giving an improvement in sensitivity of up to a factor of 5 for sustained observations . in europe , the evn , especially when combined with emerlin , will give improved surface brightness sensitivity , and evlbi will offer new target - of - opportunity possibilities , with real - time bandwidths of 1 gbps . continued radio observations , combined with the unprecedented glast all sky monitoring , may give insight into the location and mechanism of gamma - ray outbursts . in 2012 , vsop-2 will give an order of magnitude improvement in both sensitivity and resolution over halca along with polarization capability .
this will allow the transverse structures and magnetic field orientation in the m87 and other jets , to be studied in much better detail than previously possible .
in particular , vsop-2 will better define the structure and evolution of jet - deflection / interaction surfaces and the role kelvin - helmholtz instabilities in sources such as 3c 111 and 3c 279 .
however , frequent time sampling and good u , v coverage will be needed to fully exploit these capabilities .
the national radio astronomy observatory is operated by associated universities , inc .
, under a cooperative agreement with the national science foundation .
yyk is currently a research fellow of the alexander von humboldt foundation .
dch was partially supported by an award from the research corporation .
mk was supported in part through a stipend from the international max planck research school for radio astronomy at the university of bonn and , in part , by a nasa postdoctoral program fellowship appointment at the goddard space flight center .
the mojave project is supported under national science foundation grant ast-0406923 .
|
the major multi - epoch vlba programs are described and discussed in terms of relativistic beaming models . broadly speaking the observed kinematics are consistent with models having a parent population which is only mildly relativistic but with lorentz factors extending up to about 30 .
while the collimation and acceleration appears to mainly occur close to the central engine , there is evidence of accelerations up to 1 kpc downstream .
generally the motion appears to be linear , but in some sources the motion follows a curved trajectory . in other sources ,
successive features appear to be ejected in different directions possibly the result of a precessing nozzle .
the launch of glast in 2008 will offer new opportunities to study the relation between radio and gamma ray activity , and possibly to locate the source of the gamma ray emission .
vsop-2 will give enhanced resolution and will facilitate the study of the two - dimensional structure of relativistic jets , while radioastron will provide unprecedented resolution to study the fine scale structure of the jet base .
|
there are in the literature several global descriptions of variable stars .
we can mention four books , one by richter , wenzel and hoffmeister ( 1985 , out of print ) , one by sterken and jaschek @xcite , one by percy ( @xcite ) and one , soon to come , by aerts et al .
there have also been two excellent and detailed reviews by gautschy and saio ( @xcite , @xcite ) about a decade ago . with the advent of charge - coupled devices ( ccds ) and large scale surveys like the optical gravitational lensing experiment ( ogle ) ,
the massive compact halo object project ( macho ) or the `` exprience pour la recherche dobjets sombres '' ( eros ) , the subject of variable star studies is now under a remarkable expansion . as we will see
the expansion will continue in the coming decade because there are many future projects which will sample frequently large regions of the sky .
in other words there will be a larger number ( hundreds of million to billion ) of objects being measured repeatedly .
there is also a drastic improvement in the photometric quality with certain projects , exploring new regions of the variability levels . with this expansion of observational facilities
, new variable types have been discovered over the past 10 years , and important advances have been made in the understanding of pulsation theory . in this review ,
we first present the different classes of variable stars in sect .
[ sect : tree ] , and mention in sect .
[ sect : projects ] the current and future projects that benefit to the field of variability research .
we then present several observational diagrams .
the classical hertzprung - russel ( hr ) diagram is tackled in sect .
[ sect : hr ] , while other observational diagrams are addressed in sect .
[ sect : other diagrams ] .
a brief overview of the progresses recorded in the field during the last decade is presented in sect .
[ sect : progress ] .
a note on some points pertaining to stellar evolution models is given in sect .
[ sect : stellar evolution ] .
the paper ends with conclusions in sect .
[ sect : conclusions ] .
the _ general catalogue of variable stars _ ( gcvs ) lists more than 100 types and subtypes of variable stars .
the variability properties of the light curves are very diverse , ranging from extremely regular pulsations like cepheids to unique events such as cataclysmic phenomena or supernovae explosions , with times scales from the order of milliseconds for gamma ray bursts to centuries for secular evolution , and with amplitude variations from parts per million for solar - like oscillations to many orders of magnitudes for the most energetic phenomena in the universe like hypernovae . before covering the variable stars across the hr diagram , we attempt to group , in figure [ fig : vartree ] , the different types of variable sources that are present in the universe according to the physical phenomena at the origin of their variability .
four division levels are introduced .
first , the `` classical '' division between _ extrinsic _ and _ intrinsic _ variables .
the former group identifies objects whose light curve variability results from a geometrical effect , either due to the position of the object relative to the observer or due to rotation .
the latter group , on the other hand , results from physical changes occurring in the object itself .
the second level of division concerns the type of object considered , being either asteroids , stars , or galaxies .
we should indeed not forget that most asteroids and active galactic nuclei ( agn ) are observed with variable light intensities .
the third level identifies the phenomenon at the origin of the variability .
rotation , eclipses by a companion or by a foreground object , and microlensing effects are the phenomena considered for extrinsic variability , while intrinsic variable objects include eruptive stars , cataclysmic systems , stars displaying secular evolution and , last but not least , pulsating stars . finally ,
the fourth division groups objects of either similar physical nature or of similar photometric behaviour .
two remarks must be given here .
first , as in biology , the division into distinct classes is very useful , but has also its limits .
the transition between one class to another may be continuous in some cases ( e.g. sarv , l , sr , m ) , leading to difficulties to attribute an object to one class or another if its properties overlap those of two contiguous classes of objects .
second , the importance of a good nomenclature must be stressed to avoid confusion .
it is of course difficult to find a good name for a newly discovered class of objects , as witnessed by the confusion found in the literature about some classes of variable stars ( some examples are shown in sect.[sect : progress ] ) .
it is probably time now to re - assess the conventional names given to the different known classes of variable objects .
this is certainly a difficult , though important , task for which a commission of the iau has been set up .
for the time being , we have mainly followed the nomenclature used in the _ general catalogue of variable stars _ ( gcvs ) .
this diagram is mostly oriented towards photometric variability .
obviously variable objects may have signatures in their spectra that are not necessarily detectable in their photometric light curve .
such is for example the case with doppler shifts or line profile variations , which play an important role in astero - seismological studies or planet detection .
those cases are not reflected in figure [ fig : vartree ] and are not considered in this review . variability tree .
a tentative organisation of variable objects.,width=642 ] part of this conference is devoted more specifically to asteroseismology .
therefore we list the observational properties of the most common pulsating variable stars in table [ tab : varstarchar ] .
.[tab : varstarchar ] most common pulsating variable stars and their fundamental characteristics .
the periods are expressed in seconds ( s ) , minutes ( min ) , hours ( h ) or days ( d ) .
[ cols="<,<,<,<,<",options="header " , ]
many large scale surveys have accumulated multi - epoch photometric data during the last decade .
thanks to these surveys , the number of stars of different variability types is becoming large enough to allow a significant statistical description of their properties .
these surveys include ground based projects , such as the pioneering micro - lensing surveys macho or ogle - i , ii and iii , or planetary transit surveys like the hungarian - made automated telescope survey ( hat ) , the ogle - iii survey ( which should soon become public ) , the wide angle search for planets survey ( superwasp ) , and the trans - atlantic exploanet survey ( tres ) . among the actions that can be done in order to benefit efficiently from those surveys are the release of both raw and reduced data into the public domain and the availability of a user - friendly interface that allows convenient extraction of the whole or part of the whole data set .
an example of data availability is provided by the all sky automated survey ( asas ) , which is currently continuing its data acquisition of the `` bright '' objects in the sky up to magnitude 14 - 15 .
efforts have also been developed during the last decade in multisite observations . by obtaining as continuous observations as possible , aliasing effects are reduced and precise information on all frequencies of the variable objects can be obtained .
projects such as the delta scuti network project ( dsn ) and the whole earth telescope project ( wet ) have taken this multi - site option .
variable types that have mostly benefited from such programs include compact objects , @xmath0 scuti stars and @xmath1 cephei stars .
obviously , those programs can only focus on a limited number of objects .
several plans for future ground based facilities dedicated to large scale surveys will provide unprecedented material to improve our knowledge of stellar variability for a large number of objects .
those include the large synoptic survey telescope ( lsst , 8.4-meter , completion estimated for 2013 ) and the panoramic survey telescope & rapid response system ( panstarrs , starting now a test phase and scientific studies lasting for 3.5 years ) .
one project which deserves full attention is the stellar oscillations network group ( song ) multi - site project which will perform precise doppler - velocity observations for bright objects .
the antarctica site has also been under focus and promises great potential .
there are many projects , such as the seismic interferometer aiming to measure oscillations in the interior of stars ( siamois ) , the international concordia explorer telescope ( ice - t ) , the antarctica search for transiting extrasolar planets ( astep ) , and the international robotic antarctic infrared telescope ( irait ) projects . from space ,
the past missions such as the high precision parallax collecting satellite mission ( hipparcos ; a survey type mission ) or the wide field infrared explorer mission ( wire ; a pointing type mission with extremely precise photometry ) have proved the advantages of operating above the atmosphere .
reports on the on - going microvariability and oscillations of stars mission ( most ) have been shown during this meeting , as well as some remarkable preliminary results of the convection , rotation and planetary transits mission ( corot ) that was launched at the end of last year ( 2006 ) . despite the importance of those types of mission dedicated to the continuous monitoring of the sky with a high degree of photometric precision ,
it is surprising to see , retrospectively , how the realization of such missions may be uncertain on a time span of 8 years @xcite : some missions that were foreseen to be cancelled may finally be launched ( such as the nasa kepler mission , now planed to be launched in february 2009 ) , while other missions that were approved and seemed to lie on firm grounds were eventually cancelled ( such as the esa eddington mission ) .
lrr + & _ 120,000 _ & + variables & 11,597 & 9.66% + non - periodic / unsolved & 5,542 & 4.62% + not investigated & 3,343 & 2.79% + periodic & 2,712 & 2.26% + delta scuti & 186 & 0.16% + rr lyrae & 190 & 0.16% + cepheid & 273 & 0.23% + mira & 200 & 0.17% + other pulsating & 300 & 0.25% + eclisping binaries & 917 & 0.76% + others & 646 & 0.54% + + & _ 140,000 _ & + variables & 4,000 & 2.86% + as of mid-2005 & _ 7,300,000 _ & + variables & 24,936 & 0.34% + eclipsing binaries & 5,449 & 0.07% + mira & 1,502 & 0.02% + regular pulsating & 2,416 & 0.02% + other ( misc . ) & 15,569 & 0.21% + + & _ 500,000 _ & + variables & 700 & 0.14% + + & _ 2,000,000 _ & + @xmath2 & 400,000 & + variables & 2,800 & 0.70% + periodic & 1,900 & 0.48% + pulsating & 250 & 0.06% + misc . & 900 & 0.23% + eclipsing binaries & 1,650 & 0.41% + + & _ 30,000,000 _ & + variables & 200,000 & 0.67% + & _ 9,000,000 _ & + variables & 68,000 & 0.76% + rr lyrae ( smc ) & 571 & 0.006% + cepheids & 3382 & 0.034% + eclipsing binaries ( smc ) & 1459 & 0.016% + qso ( behind smc ) & @xmath3 6 & + we must note that some surveys that are designed for other fields of astronomy may nevertheless also contribute to the field of variability studies . the sloan digital sky survey ( sdss ) , for example ,
whose primary goal was cosmology , allowed to perform variability studies on rr lyrae stars @xcite .
some regions of the sky were covered by overlapping fields during the survey , leading to multi - epoch observations .
the sdss also led to the discovery of asteroids through the detection of their relative motion in successive exposures @xcite .
in addition , blake et al .
@xcite succeeded to find a very low mass eclipsing binary system in a time series containing only 30 epochs .
as an example of a future survey in space , we mention the gaia mission , a cornerstone esa mission with a foreseen launch date in 2011 .
it will observe one billion stars , mostly in our galaxy , and provide astrometric , photometric , spectrophotometric and radial velocity multi - epoch observations .
the requirements for the astrometric precision ( parrallax error for a 5 year mission ) mention `` better than 24 micro arcsec for a g2 star at mag @xmath4 '' .
the satellite will record a mean of about 80 measurements per object spread over 5 years .
the photometric precision is estimated to about 1 - 2 millimag per measurement for objects with a magnitude up to about @xmath5 .
the satellite will have the opportunity to detect a large number of new variables , including short time pulsators of several tens of seconds .
if gaia is operating at the precision level requested in the specifications , it will provide , among other results , hr diagrams with an unprecedented level of information .
the preparation of a mission is also an opportunity to develop new tools . in the case of gaia , for example , a model of the galaxy
is being developed that will provide , as a by - product , a realistic representation of the distribution of variable objects in the sky .
a better definition of the classes of variable objects than the one given in table [ tab : varstarchar ] must be established . to this end
, we are assembling data and defining probability distributions for the variability properties of different classes of variable objects in various observational diagrams ( see sects .
[ sect : hr ] and [ sect : other diagrams ] ) .
the goal is to predict the number and types of variables as well as their properties present in a given population .
the number of variables detected in a survey depends on the observed stellar population(s ) , on the covered range of magnitudes , on the length and the sampling law of the observation , on the overall instrument precision , and on the data processing methods used for variability analysis .
the number of variable objects detected in a few surveys is presented in table [ tab : numberofvariables ] .
the largest fraction of variables is so far obtained by the hipparcos mission , with 9.7% of the 120,000 observed stars being variable .
fraction of variable across the hr diagram from hipparcos catalogue @xcite.,width=453 ] hr diagram with colour coding for amplitudes .
the data are taken from the annexes `` unsolved variables '' and `` periodic variables '' in the hipparcos catalogue.,width=453 ] hr diagram with colour coding for periods .
the data are taken from the annex `` periodic variables '' in the hipparcos catalogue .
some types of pulsating and rotation induced variables are shown in the figure.,width=453 ] the hr diagram is often expressed in terms of the effective surface temperature and absolute bolometric luminosity , which are the physical quantities intrinsic of stars .
the observed quantities , however , are usually the apparent magnitude and a colour index .
there are several caveats when converting observable to intrinsic quantities .
the transformation of the apparent magnitude into an absolute magnitude requires the knowledge of the distance to the object , which can be obtained either with the parallax for objects close enough to the earth , with some calibration procedure ( cepheids for example ) , or with the knowledge of the distance to the group of stars to which the object belongs , if any ( this is the case for clusters , the magellanic clouds , or galaxies ) .
in addition , since the apparent magnitude is measured in a given band , a bolometric correction must be applied to infer the bolometric magnitude .
a most often used hr diagram is that of the absolute v magnitude @xmath6 versus the b - v colour
. we must keep in mind that the morphology obtained in an hr diagram based on observable quantities varies - sometimes dramatically - depending on the spectral bands that are used for the colour and/or apparent luminosity .
we must also keep in mind that some stars may display large variations in @xmath6 , but tiny ones in the bolometric magnitude .
the interstellar extinction , metallicity , duplicity and rotational effects may also affect the morphology of the hr diagram .
these effects can be far from being negligeable .
the interpretation of an hr diagram may thus be delicate .
a third dimension can be added to the colour - magnitude hr diagram by using colours to scale a third parameter . for variable stars
the third parameter can be the fraction of variable stars , the amplitude of variability , or its period .
for hipparcos data , such diagrams have been published in refs .
@xcite , @xcite and @xcite .
the hipparcos diagrams within these last two references were used with the survey part of hipparcos catalogue ( in order to avoid biases from the selected sample of hipparcos ) and excluding the eclipsing binaries . for the geneva photometry database , they can be found in @xcite .
the diagrams presented in the hipparcos catalogue ( @xcite ) were obtained using the absolute magnitude , while the other papers presented the diagrams in terms of the spectral and the luminosity classes .
the distribution of the fraction of variable stars from the hipparcos catalogue is shown in figure [ fig : hrfofv ] . that of the amplitudes and periods are displayed in figures [ fig : hramp ] and [ fig : hrper ] , respectively .
hr diagram with eclipsing binaries .
the data are taken from the annex `` periodic variables '' in the hipparcos catalogue .
crosses represent ea type binaries , open circles eb , filled squares ew.,width=453 ] the location of eclipsing binaries in the hr diagram is presented in figure [ fig : hreb ] .
the three classical types of eclipsing binaries are presented : ea algol type , eb @xmath1 lyr type , and ew w uma type .
note that other classifications of eclipsing binaries , probably more adequate to describe the physical properties of those systems , are available in the literature ( e.g. @xcite , @xcite ) .
it is interesting to note that ew types are confined in a smaller region in the hr diagram than ea and eb types .
hr diagrams with smc ogle - ii cepheids.,width=453 ] we can have a closer look at the cepheids in the smc @xcite .
ogle - ii provided a list of 2167 cepheids , that we can plot in a hr diagram .
this is shown in figure [ fig : hrcep ] , where specific colours have been assigned for given ranges of pulsation periods .
the periods are very clearly distributed , the longer ones being at higher luminosities and redder colours .
this is expected from the period luminosity relation .
the motion of a cepheid in the hr diagram is presented in figure [ fig : hrmotion ] .
motion of the position of the cepheid u nor in the hr diagram.,width=604 ]
many diagrams other than the hr diagram presented in the previous section can be constructed based on quantities that can be readily derived from observational data .
we mention in this section some of them .
the list is not exhaustive , but gives a flavor of the importance of those diagrams in analyzing the properties of variable stars .
period - amplitude diagram ( logarithmic scale ) for hipparcos variable stars.,width=604 ] when the absolute magnitude can not be obtained , we can still work on a colour - colour diagram .
such diagrams , however , are more degenerate than the classical hr diagram .
there is for example an overlap between the region covered by the main sequence stars and the region covered by giant stars .
nice examples of such colour - colour diagrams are given in @xcite for halo stars observed by the sdss .
in these diagrams , the authors provide information on various variability properties in the third dimension .
in addition to the colours and luminosities , there are two other quantities that can easily be derived from the observation of variable stars .
those are the period(s ) and the amplitude(s ) of variability measured from the light curves in one or several photometric bands .
these quantities are intrinsic properties of the star and are independent of the distance .
the amplitude of variability may , however , vary a lot depending on the photometric band considered .
the phase of the periodic signal may also vary with the photometric band .
the period - amplitude diagram constructed from the hipparcos data of variable stars is shown in figure [ fig : p - a ]
. expressed in logarithms of the periods and amplitudes , the diagram shows very clearly distinct regions of clumped data that correspond to different types of variable stars .
the combined usage of period and amplitude of variability is thus an efficient way to separate variability classes .
there are , however , regions that contain variable stars of different classes .
the region covered by @xmath1 cep stars , for example , overlaps either with the region of slowly pulsating b stars , or with that of @xmath0 scuti stars .
the region around @xmath7 and @xmath8 ( @xmath9 and @xmath10 ) is another example of a region mixing several types of variable stars , gathering in this case sx phoenicis , rr lyrae and @xmath0 scuti stars .
the diagram shown in figure [ fig : p - a ] is based on the variability type assignments listed in the `` catalogue of variable stars '' of the hipparcos mission at the exception of the @xmath11 dor stars , which were extracted from simbad .
it must be noted that there has been no specific work devoted to the global systematic classification of variability types in the hipparcos catalogue , at the exception for the distinction of eclipsing binaries ( because of the doubling of the period ) , the rv tauri stars ( for the same reason ) , and for the b stars .
moreover , some variability class assignments were made based on information about the star taken from the literature , i.e outside the hipparcos data
. it would be interesting to reanalyze the hipparcos data with new classification methods based on machine learning techniques .
such methods are under development in the framework of the preparation of the data analysis pipeline foreseen for the gaia mission .
preliminary results applied to the hipparcos data are published in @xcite .
probability distribution in the period - amplitude diagram for galactic cepheids ( left figure , from the fernie catalogue of the david dunlap observatory ) and for rr lyrae in the large magellanic cloud ( right figure , from the ogle - ii data base ; the clump on the lower left identifies the rrc type while the elongated branch is the rrab type ) .
, width=604 ] a probability distribution diagram can also be constructed for a given variability class , if the number of stars pertaining to the class is high enough .
such a probability distribution in the period - amplitude diagram is shown in figure [ fig : probdistr ] for galactic cepheids ( left figure ) and for rr lyrae in the large magellanic cloud ( right figure ) .
we see that the large surveys currently available allow now to obtain a statistical description of given populations of stars that is significant .
caution must be taken , however , with the observational biases introduced , for example by the type of population observed and/or by the light curve sampling .
they can lead to biases in the distributions that can render their interpretation tricky . taking this into consideration
, we can expect interesting and challenging results to be obtained in the near future from the statistical analysis of the data collected by large surveys , especially when comparing stellar populations differing in well defined parameters such as the metallicity .
finally , we have to mention that , similarly to what may happen in the hr diagram for variable stars ( see end of sect .
[ sect : hr ] ) , a point representative of a given star may move in the period - amplitude diagram , depending on the time at which the amplitude and/or period are derived .
the amplitude of variability of an rr lyrae may change in time due to the blazhko effect . despite being known since almost a century , this phenomenon of amplitude and/or phase modulation ( on a time scale five to ten times longer than the rr lyrae period )
still remains a puzzle for theoreticians .
the period of a variable star may also change with time , for many different reasons .
a usual way to analyse the period variability is to plot as a function of time the difference between the observed period and the one expected in case of no period evolution .
such diagrams are called o - c ( observed - calculated ) diagrams . the slow evolution of periods observed in cepheids , for example , has been analyzed in this way by @xcite , by comparing the data of ogle and asas taken during the years 1997 - 2000 and harvard data taken between 1910 - 1950 .
we refer to the book of sterken @xcite which gathers the proceedings of a conference entirely devoted to the analysis of periods using o - c diagrams .
such period changes may lead to motions in the period - amplitude diagram .
a rather recent breakthrough has been brought thanks to interferometry techniques .
the angular diameters and even their change caused by pulsation have been directly measured with interferometry .
this allows to construct a new observational diagram , which is the period - radius diagram , see @xcite .
furthermore , the period luminosity relation can be derived avoiding the determination of the temperature .
the field of variable stars and astero - seismology has been very active during the last decade , and substantial progresses have been achieved for several classes of variables . in this section ,
we highlight some of them that cover different regions in the hr diagram .
the choice of variability classes is necessarily arbitrary , and by no means exhaustive .
for example we do not mention @xmath12 cyg stars , luminous blue variables ( lbvs ) , or white dwarfs .
the purpose is to give a flavour of the on - going research activity .
we also note that some recent results found in the literature are still matter of debate , which after all is indicative of an actively evolving field .
we start with some types of variables found on the main sequence ( sect .
6.1 - 6.3 ) and consider red giants in sect . 6.4
. some late phases of evolution are addressed in sect .
6.5 to 6.7 .
evidence of solar - like oscillations was claimed by @xcite in @xmath13 boo , but the detection has been controversial .
the first strong identification of individual frequencies from such solar - like oscillations in a star other than the sun was reported by @xcite in @xmath12 cen a. this gave a real kick to the field of asteroseimology .
the clear detection was made possible thanks to the improvement brought , within the planet search program , to the coralie instrument in measuring radial velocities . for @xmath12 cen
a , amplitudes of 35 cm / s could be measured . since then , the detection of solar - like oscillations has been reported for several other stars .
we refer to the talk of eggenberger in this volume .
three types of variables are known among the mains sequence b stars : the @xmath1 cephei stars , the slowly pulsating b ( spb ) stars , and the be stars .
the variability observed in @xmath1 cephei and spb stars is due to pulsation triggered by the @xmath14 mechanism on the iron - group elements .
main sequence b stars displaying the pulsational characteristics of both @xmath1 cephei and spb variables are also known .
be stars , on the other hand , are defined by the presence of emission line(s ) in their spectra arising from a circumstellar disk , and are characterized by irregular light curve variability on time scales from weeks to years related to the disk .
but since the be phenomenon is transient , those stars can also exhibit periodic variability due to the same @xmath14 mechanism observed in other @xmath1 cephei and spb stars .
be stars are fast rotators .
the study of @xmath1 cephei stars has witnessed great progress during the last decade thanks to sustained efforts to conduct multisite observations .
the astero - seismological potential of @xmath1 cephei stars was re - emphasized when @xcite reported to be able to probe the inner regions of the star hd 129929 based on the analysis of photometric data gathered during 21 years .
they claimed to provide evidence for the occurrence of core convective overshooting , with an overshooting parameter between 5 and 15% of the pressure scale height , as well as a hint for nonrigid rotation ( @xcite ) .
multisite observation campaigns led thereafter to the publication of detailed astero - seismology analysis for @xmath15 eridani ( @xcite , @xcite ) and several other @xmath1 cephei stars ( e.g. @xcite , @xcite ) .
they led to the detection of several pulsating modes , and allowed to put constraints on the global stellar parameters as well as on the core overshooting parameter of the observed b stars .
the number of pulsating main sequence b stars is expected to decrease in stellar populations of decreasing metallicities , which seems confirmed by the observation of the number of pulsating b stars in several young open clusters of different metallicities @xcite .
this is due to the fact that the efficiency of the @xmath14 mechanism on iron - group elements of course depends on the abundances of those elements .
models predict the vanish of the @xmath1 cephei instability strip at @xmath16 , while the spb instability strip remains noticeable down to @xmath17 .
given the low metallicities of the magellanic clouds , they provide interesting laboratories to test these predictions .
using the data from the large scale mircolensing surveys ogle and macho , @xmath1 cephei and spb stars were discovered in the large magellanic clouds ( @xcite , @xcite ) .
but the metallicities of the stellar populations in the lmc span a large range , from @xmath18 to 0.01 ( @xcite ) , making it difficult to check the number of pulsating b stars as a function of metallicity .
more convincing would be to check the presence ( or absence ) of variable b stars in the smc .
such a study has been recently undertaken by @xcite .
they found nine b stars showing short - period variability .
surprisingly one of them is a @xmath1 cephei variable , the eight remaining ones being spb stars .
these authors also found three be stars with pulsation periods shorter than 0.5 days , and suggest those to also be @xmath1 cepheid stars . if these result are confirmed , and if the metallicity of those proposed @xmath1 cephei stars are indeed similar to the representative metallicitiy of the smc , i.e. @xmath19 , then they provide a new challenge for stellar models .
rapidly oscillating ap stars are a subclass of chemically peculiar a stars which display photometric pulsations ( non - radial ) with periods between 5 and 20 minutes .
the small amplitudes of their variability , typically less than a few millimagnitudes , partly explain their relatively late discovery @xcite .
recent important progress was achieved in the study of those stars by analyzing the temporal variability of high - resolution spectral lines of different chemical elements .
those lines are formed at different depth in the atmosphere , depending on the chemical element .
the pulsational characteristics result from the short vertical length of pulsation waves combined to the chemical stratification characteristic of the atmosphere of chemically peculiar stars .
therefore , the analysis of the spectral lines variability provides information on the pulsation properties at different atmospheric depths , allowing to perform tomography of the atmosphere of those stars .
we refer to the reviews by @xcite and @xcite for further details on the observation of pulsations in roap stars .
several classes of variables are known since long among red giants .
initially , those were classified as miras , semi - regular variables , and irregulars . the term long period variable is also used .
because of the high amplitude variations of miras , the variability of those giants could be studied since very early . a period luminosity relation ( pl ;
see review by @xcite ) was obtained at the beginning of the twentieth century @xcite . since then and until about a decade ago
, progress on the observational ground has led to a tightened relation , to the discovery of a second pl relation for semi - regular variables , parallel to that of miras but shifted to shorter periods ( @xcite ) , and to the better knowledge that not only the variability amplitudes increase as the surface temperature of the red giant decreases ( @xcite , @xcite ) but also that _ all _ red giants with spectral types later than early k are variable , with a minimum variability level that increases with increasing spectral type ( @xcite , @xcite ) .
the availability of large surveys in the last decade , such as macho and ogle , opened a new era in the observational study of red giant variables .
it is remarkable to see how the statistical analysis of tens of thousands of giants in the magellanic clouds led to the identification , so far , of fourteen distinct relations in the pl plane ( @xcite and references therein ) .
the richness of information in those diagrams is such that it allows to split the near infrared pl sequences in the period - optical wesenheit index plane into two separate ridges that seem to correspond to the spectral division of o - rich and c - rich agb stars ( @xcite ) .
small amplitude variability has been highlighted among the ogle red giants , baptised ogle small amplitude red giants ( osargs ; @xcite ) , the nature of which is still unclear .
the ogle observations of the magellanic clouds also provide input to study long secondary period variables ( @xcite ) . on the theoretical ground ,
the physical origin of red giant variability remains a subject of active research .
any quantitative analysis of variability in those stars requires a time - dependent description of convection , the coupling of which with pulsation is an important ingredient in providing reliable predictions .
we discuss this point , and its results for the understanding of red giant variables , in section [ sect : stellar evolution ] .
pulsating b sub - dwarfs have been discovered in the last decade ( @xcite ; see @xcite for an observational review ) , and have since become a nice example of an astero - seismological field where great progress has been achieved in a short time .
this new field has benefited from observational efforts that include long multi - site campaigns , both photometric and spectroscopic , as well as from progresses on the side of stellar modelling . a complete astero - seismological description of a dozen of sdb stars , leading to the precise estimation of the stellar properties ( surface gravity , surface temperature , stellar mass , envelope mass , radius , luminosity , rotational period , ... ) ,
are available to date .
we refer to the excellent review of @xcite , in which an example of an end - to - end astero - seismological exercise can also be found .
there are about 70 pulsating sdb stars known so far , comprising two groups .
one group consists in rapidly pulsating sdb stars , on time scales of few minutes , commonly known as ec 14026 variables .
they are pulsating in p - modes .
the second group of pulsating sdb stars , the so called _ betsy _ variables , have periods of about 0.5 to 2.5 hours .
they are g - mode pulsators .
the two groups define distinct regions in the @xmath20 plane .
the ec 14026 stars occupy the high temperature side at @xmath21 k , while the _ betsy _ stars are found below this temperature .
remarkably , two stars have been discovered at the boundary separating the two regions , that exhibit both p - mode and g - mode pulsations ( @xcite , @xcite , @xcite ) .
the sdb stars are compact helium stars with masses around 0.5 m@xmath22 , that lie at the high temperature extension of the horizontal branch in the hr diagram .
they are therefore also called extreme horizontal branch stars .
while the evolutionary scenario that leads to their formation is not yet known , the options being single star or binary evolutions , their structure is rather well defined .
they are burning he in their core , but have only a thin h - rich envelope above it .
therefore , they will not be able to expand as an agb after core he exhaustion , but will directly move left in the hr diagram to the white dwarf stage .
the pulsations observed in those stars originate from the @xmath14 mechanism on iron group elements .
for the mechanism to become efficient , however , diffusion processes must act in the envelope to increase the relative abundance of those heavy elements in their partial ionization zone .
this is indeed predicted by evolutionary models of those stars .
the existence of pulsating sdb stars was actually predicted at the same time than , but independently of , the discovery of ec 14026 - 2647 , the first prototype of ec 14026 stars @xcite .
it is the hope of the experts in the field to reach the point where enough statistics are reached in the properties of sdb stars to provide a stronger clue on their evolutionary origin .
the region in the hr diagram occupied by post - agb stars comprises several types of variable stars .
they include both h - rich stars such as rv tau stars , and h - depleted stars such as r crb stars at the cool side of the hr diagram or pv tel and extreme he stars at the hot side of that diagram ( the case of gw vir stars is presented separately in the next section ) .
note that the distinction between rv tau and population ii cepheids is not always evident based solely on their pulsational characteristics ( see the review of @xcite , for example ) . from an observation point of view , most rv tau stars show infrared excesses , in contrast to pop
. ii cepheids , which is consistent with rv tau stars being post - agb stars
. there has not been fundamental advances on the study of those variables in the last decade and the field remains to be explored more deeply in future , most probably with the help of large scale surveys .
some points are however worth mentioning .
rv tau stars are post agb stars that become pulsationally unstable as they cross the classical instability strip during their motion to the hot side of the hr diagram . in a recent study , @xcite have been able to convincingly illustrate this fact by analyzing the pulsation properties of 17 post - agb stars from the asas and the northern sky variability survey ( nsvs ) .
the stars observed within the instability strip have a well defined periodicity , while stars lying at either edge of the strip are multiply periodic or possibly irregular .
h - deficient post - agb stars are also expected to be studied in more detail during the coming years . the majority of them are thought to result from a last thermal instability that occurs in their helium - burning shell , either when they reach the hot side of the hr diagram or when they are already descending the white dwarf track .
this late thermal pulse , as it is called , expands the thin envelope and turns the hot post - agb star back to a red giant .
concomitantly , nucleosynthesis products of both the former hydrogen and helium burning shells are mixed into the envelope , resulting in a h - deficient envelope .
while this scenario can explain the presence of most of the h - deficient stars , a fraction of them are almost exclusively dominated by pure helium .
the formation of those extremely hot helium - rich stars may require a scenario involving the merge of two white dwarfs .
they would also be the progeny of rcrb stars .
we refer to the short conference article of @xcite for further references on this subject .
finally , let us mention that the stellar evolutionary time scales of those post - agb stars are relatively short .
they could lead to secular evolution that could be detectable on human time scales ( e.g. @xcite ) , such as witnessed by the well known stars fg sagittae and the sakurai object .
the availability of surveys spanning several years may provide new opportunities to detect such objects .
needless to say , they would offer unique ways to test stellar evolution models .
there is a particular class of post - agb stars that has attracted the attention of both observers and theoreticians in the past decade , some of those stars being among the best astero - seismically studied .
those are h - deficient post - agb stars on their way to the white dwarf cooling sequence .
those h - depleted atmosphere stars include the [ wc ] wolf - rayet central stars of planetary nebula , the pg 1159 stars that exhibit absorption lines of highly ionized he , c and o , and the do he - dominated atmosphere white dwarfs .
those types of stars are evolutionary connected through the sequence [ wc ] @xmath23 pg 1159 @xmath23 do .
we refer to @xcite for a review on those objects .
a good fraction of those h - deficient stars are pulsating .
various variability types are encountered in the literature , such as the gw vir stars , usually associated with pg 1159 variables , the pnnv , initially thought as the planetary nebula nucleus variables , and the dov for the high surface gravity gw vir stars .
as mentionned by @xcite , the terms pnnv and dov are misleading and do not identify separate physical mechanisms for the pulsation .
therefore , we follow @xcite and use the name gw vir to qualify all variable h - deficient stars that lie in that part of the h - r diagram .
they all have a common pulsation mechanism , which is the @xmath14-mechanism associated with the partial ionization of carbon and oxygen ( k - shell electrons ) .
it was @xcite who first suggested the existence of a new instability strip based on the @xmath14 mechanism for partial ionization of c and o to explain the @xmath3539 s and @xmath3460 s pulsations detected in the newly discovered @xcite pg 1159 - 035 variable ( = gw vir in the naming of variable stars ) .
pg 1159 - 035 would pulsate in low - order g - modes .
the high effective temperature of this star ( 140,000 k ) further suggested that the modes are non - radial ( @xcite ) . since the discovery of the gw vir instability strip , controversial results were presented in the literature concerning the efficiency of c and o to induce the pulsations in the presence of he .
the latter acts as a poison that smoothes the opacity gradients that would otherwise be obtained by the partial ionization of c and o ( @xcite ) .
some authors confirm this difficulty ( e.g. @xcite , @xcite ) , others obtain satisfactory results ( @xcite , @xcite ) .
the latest , comprehensive , analysis of gw vir models seems to confirm the c - o ionization origin of the pulsation , and shows that the extent of the instability domain in the log @xmath24-@xmath25 plane is a strong function of the c and o abundance in their envelopes @xcite .
the presence of pulsationally stable and unstable objects observed in that part of the h - r diagram is explained in terms of different chemical compositions in the envelope of those h - deficient stars .
the interpretation of the observed properties of variable and pulsating stars relies heavily on stellar models , whether to derive the global stellar parameters or to identify pulsation modes .
our understanding of the different regions covered by variable stars in the hr diagram is also based on stellar evolution model predictions . in this section ,
we highlight some development in this area during the last decade that has enabled us to better understand the variability properties across the hr diagram .
it is by far not exhaustive . among the important processes not covered here
are rotation and chemical diffusion processes .
convection is one of the physical processes that is difficult to be properly taken into account in stellar evolution models .
being an essentially three dimensional problem , convection can only be included in one dimensional evolution models with a great deal of simplification .
while the classical parameterized mixing length theory is a convenient prescription to be numerically included in stellar evolution codes , it can not describe satisfactorily the coupling between convection and pulsation .
consequences of such shortcomings include , for example , the failure of standard models to predict the correct location of the red edges of instability strips , the determination of which requires a non - adiabatic treatment of the interaction between convection and pulsation .
new time dependent models of convection have been developed in the last decade to overcome these shortcomings .
successful results were obtained for rr lyrae and cepheids ( @xcite ) , for @xmath0 scuti stars ( @xcite , @xcite , @xcite ) , for @xmath11 dor stars ( @xcite , @xcite ) , for pre - main sequence stars ( @xcite ) , and for red giants ( @xcite ) .
the coupling of convection and pulsation is of particular importance for red giants , which are characterized by deep convective envelopes .
for example , @xcite convincingly claim to be able to predict , using appropriate time dependent models of convection , both the variability of red giants and the non - variability of stars lying , in the hr diagram , between the red giant branches and the classical instability strip of cepheids / rr lyrae .
they identify a red giant instability strip in the lower temperature side of the hr diagram which goes along the sequences of the red giant branch and the asymptotic giant branch .
the excitation mechanism of that instability strip , they say , is related to the turbulent pressure of the convective region .
their models further predict that red giants of low luminosity pulsate at high order overtones , the fundamental to second overtone being stable , while the instability moves to lower order modes when considering red giants of higher luminosity and lower effective temperature . at the high luminosity end of the strip ,
red giants pulsate only in the fundamental and first overtone modes .
if confirmed , this scenario provides a nice explanation of many of the patterns observed in the variability zoo of red giants .
it also raises new issues .
we refer to the paper of @xcite for further explanations and discussions .
the chemical abundance mixture is another parameter of stellar evolution that impacts model predictions . for a given metallicity , the relative abundance of elements heavier than helium
is usually scaled to that of the sun , whose abundance was inferred , until recently , using 1d models of the solar atmosphere .
however , new 3d hydrodynamical models have led to a major revision of those solar abundances , resulting in significantly lower carbon , nitrogen , oxygen and neon abundances ( @xcite , @xcite ) .
the new solar metallicity is now @xmath26 , almost a factor of two lower than previously adopted . while these revised abundances bring the solar abundance in agreement with that measured in the solar neighborhood ( @xcite ) , it destroys the remarkable agreement that was hitherto found between the sound speed profile observed by inverting helio - seismology data and that predicted by models ( @xcite ) .
the adoption of a new solar abundance mixture asks for a revision of the iron mass fraction derived at other metallicities as well
. this may be particularly important for @xmath1 cephei and spb stars that pulsate due to the @xmath14 mechanism acting on the partial ionization zones of the iron - group elements . and ,
indeed , a revision of the instability strip for those pulsating stars seems necessary .
we refer to @xcite for further discussion on this point .
there is a flood of data being and becoming available thanks to a large number of ground based and space projects . in many cases
, there is a spectacular improvement in the precision ( ground based radial velocity measurements or space - based photometric measurements ) , and new observable quantities such as the direct measurement of stellar radii with interferometry become possible .
we witness many advances in the knowledge of pulsation , but many things still remain to be understood . we could paraphrase the cosmologist mark kamionkovski and conclude with conviction : `` now is the time to be an asteroseismologist '' .
we would like to thank the organiser of the meeting , laurent gizon .
our thanks go also to mihly vradi , who helped us in some searches from the literature , and to jan cuypers for contributing to table [ tab : numberofvariables ] .
9 g. fontaine , p. brassard , s. charpinet , e.m .
green , p. chayer , s. randall , b. dorman 2006 , proceedings of soho 18/gong 2006/helas i , _ beyond the spherical sun _ , eds . k. fletcher , m. thompson , esa sp-624
|
an overview of pulsating variable stars across the observational hertzprung - russel ( hr ) diagram is presented , together with a summary of their global properties .
the hr diagram is presented with a third colour - coded dimension , visualizing the fraction of variable , the amplitude of variability or the period of variability .
the distribution of variable stars in the other observational diagrams , such as the period - amplitude diagram , is also presented .
some of the progresses performed in the field of variable stars during the last decade are briefly summarized , and future projects that will improve our knowledge of variable stars are mentioned .
|
galaxy - galaxy gravitational lensing is a powerful astrophysical tool for studying the distribution of matter , including dark matter , in galaxies @xcite
. one promising application of lensing is to study and measure the properties of dark matter substructures in the lens galaxy @xcite . in recent years
, this promise has been realised in several lens systems which are thought to contain at least one dark substructure @xcite . in these studies ,
the lens was modelled as a superposition of a smooth component ( such as an elliptical power - law matter distribution ) plus a pixellized `` non - parametric '' correction term .
the estimated spatial structure of the correction term provides clues about the locations of possible substructures . in a second modelling step ,
the lens is assumed to be a smooth overall component plus a compact `` blob '' near any locations suggested by the correction term .
this kind of inference has also been done with point - like multiple images of quasars @xcite . in this paper
, we introduce an approach for solving this same problem ( inferring the number and properties of dark substructures given an image ) in a more direct fashion , by fitting a model with an unknown number of substructures directly to the image data .
however , with appropriate modifications , it may also prove useful in other galaxy - galaxy lens modelling areas .
one example is time - delay cosmography , which requires realistically flexible lens models , and reliable quantification of the uncertainties @xcite .
lens modelling has been a topic of much interest over the last two decades .
most approaches are based on bayesian inference @xcite , maximum likelihood @xcite or variations thereof .
these approaches may be categorized according to the following criteria : 1 . whether to use a simply parameterized ( e.g. a srsic profile ) ,
moderately flexible ( e.g. a mixture model , * ? ? ?
* ) or free - form ( e.g. pixellated ) model for the surface brightness profile of the source ; 2 . whether to use a simply parameterized ( e.g. singular isothermal ellipsoid [ sie ] ) or flexible model ( e.g. pixellated ) for the mass profile of the lens ; and 3 .
how to compute the results ( e.g. optimization methods , markov chain monte carlo [ mcmc ] with source parameters marginalized out analytically , or mcmc with the source parameters included ) .
recent sophisicated approaches investigating different prior assumptions and computational approaches include @xcite , @xcite , and @xcite .
each approach involves tradeoffs between convenience and realism .
simply parameterized models , such as srsic surface brightness profiles for the source , and singular isothermal ellipsoid plus external shear ( sie+@xmath0 ) models for the lens , are very convenient .
they only have a few adjustable parameters , and they capture ( to `` first order '' ) relevant prior information that we have about the source and lens profiles . of course , they are clearly simplifications , and can produce misleading results if the actual profile is very different from any member of the assumed family .
on the other hand , pixellated models for the source ( e.g. * ? ? ? * ) or the lens ( e.g. * ? ? ?
* ) can in principle represent `` any '' source surface brightness profile or lens projected density profile .
however , the prior distribution over pixel values is often chosen to be a multivariate gaussian for mathematical reasons , so that the source can be analytically marginalized out @xcite .
unfortunately , a multivariate gaussian prior over pixel intensities usually corresponds to a poor model of our prior beliefs about the source .
it assigns virtually zero prior probability to the hypothesis that the source actually looks like a galaxy , and very high prior probability to the hypothesis that the source looks like noise ( or blurred noise ) .
these priors also assign nonzero probability to negative surface brightness or density values ; in fact , the marginal prior probability that any pixel is negative is typically 0.5 .
@xcite argued that an ideal modelling approach lies somewhere between simply - parameterised and pixellated models , which is also a motivation behind the investigation of shapelets by @xcite .
one way of achieving this is with mixture models .
the source and the lens can be built up from a mixture of a moderate number ( from a few to a few hundred ) of simply parameterised components .
for the source , this allows us to incorporate prior knowledge about the local correlations ( the surface brightness at any particular point is likely to be similar to that at a nearby point ) and the fact that most of the sky is dark @xcite .
the @xcite model also allowed for multi - band data , and allowed for our expectation that the image may or may not be similar in different bands .
the present paper is similar in approach to @xcite .
the main differences are : i ) we use a simpler and faster set of basis functions ; ii ) we apply it to the lens as well as the source , allowing for the possibility of substructure ; and iii ) the implementation is based on a c++ template library developed by @xcite which allows for hierarchical priors and trans - dimensional mcmc to be implemented in a relatively straightforward manner .
however , in the present paper we do not account for multi - band data ; this will be reserved for a future contribution .
the source code for this project is available at http://www.github.com/eggplantbren/lensing2 .
if we are interested in detecting and measuring the properties of possible dark substructures in a lens galaxy , a superposition of an unknown number of `` blobs '' is the most natural model . in addition , if we want to constrain the mass function of these blobs ( e.g. * ? ? ?
* ; * ? ? ?
* ) , we will need a hierarchical model which specifies the prior probability distribution for the masses conditional on some hyperparameters . inferring the mass function of the blobs
then reduces to calculating the posterior distribution for the hyperparameters .
this is related to the general principle that we should ( ideally ) construct our inference methods so that the quantities we infer directly answer our scientific questions @xcite . here
, we are referring to the mass function of the substructures within a single lens galaxy . an additional level of hierarchy would be needed to answer questions about a sample of lens galaxies .
we now describe the details of our model and its parameterization .
the motivation for most of our modelling choices is a compromise between computational efficiency , realism , and ease of implementation .
none of the choices we have made are final in any sense ; rather , this model should be considered a proof of concept .
we encourage exploration of other choices .
since we can compute marginal likelihoods , bayesian model comparison between our choices here and any proposed alternatives should be straightforward , if a union of our hypothesis space and another can be considered a reasonable model of prior uncertainty .
the surface brightness profile of the source is assumed to be composed of a sum of a finite number of `` blobs '' , or basis functions , in order to allow some flexibility while incorporating prior information about the non - negativity of surface brightness , and the spatial correlation expected in real surface brightness profiles . for computational speed , we choose the following functional form for the surface brightness profile of a single blob centered at the origin : @xmath1 where @xmath2 , @xmath3 is the width of the blob , and @xmath4 is the total flux of the blob ( i.e. the integral of the surface brightness over the entire domain ) .
these basis functions are inverted paraboloids , which are faster to evaluate than gaussians , since they do not contain an exponential function .
in addition , the finite support means that each blob will evaluate to zero over a large fraction of the domain which confers an additional speed advantage .
more conventional choices such as gaussians are possible , as are alternative compact basis functions , perhaps inspired by the smoothed particle hydrodynamics literature @xcite .
alternative choices , such as srsic profiles , may be more realistic despite their computational cost .
this is especially likely in the case of an early - type source galaxy ( e.g. * ? ? ?
such modifications are possible without great effort using the rjobject library , and once implemented , the marginal likelihood can be compared across different choices for the source model .
if our model contains @xmath5 such blobs , positioned at @xmath6 with widths @xmath7 and total fluxes @xmath8 , the overall surface brightness profile is : @xmath9 where @xmath10 . under these assumptions ,
the source can be described in its entirety by the following parameters : @xmath11 where @xmath12 denotes the parameters for source blob @xmath13 : @xmath14 the dimensionality of the parameter space for the source depends on the minimum and maximum values of @xmath5 , which we set to 0 and 100 respectively ( more detail about priors is given in section [ sec : priors ] ) .
therefore the source is described by 1 401 parameters .
the surface mass density profile of the lens is modelled as a superposition of a singular isothermal ellipsoid plus external shear ( sie+@xmath0 ) and @xmath15 circular lens `` blobs '' .
the sie+@xmath0 is a simple and widely used lens model with analytically available deflection angles , and is intended to account for the bulk of the lensing effect due to the smooth spatial distribution of ( visible and dark ) matter in the lens galaxy .
unfortunately , along with simplicity comes a lack of realism , and generalising the smooth lens model is an important future step .
the softened power law elliptical mass density or spemd model is a popular generalization of the sie where the deflection angles can still be computed quickly using the approximations given by @xcite .
ultimately , concentric mixtures of smooth components may be the most useful for realistic inference of the density profile of the halo .
a common approach in the lensing community , that is similar in spirit , is a superposition of an elliptical power law , external shear , and a pixellated potential correction .
however , the direct use of blobs is closer to the scientific question at hand when investigating postential dark substructures .
the sie+@xmath0 has nine free parameters : the ( circularized ) einstein radius @xmath16 , axis ratio @xmath17 , central position @xmath18 , orientation angle @xmath19 , the external shear @xmath0 , and the orientation angle of the external shear , @xmath20 .
the @xmath15 lens blobs are intended to model possible dark or faint substructures in the projected mass profile of the lens .
a lensing blob with mass @xmath21 and width @xmath22 centered at the origin has the following surface mass density profile : @xmath23 which is the same as the surface brightness profile of a source blob .
the deflection angles for a single blob are : @xmath24 where @xmath25 .
these do not depend on any computationally expensive functions such as square roots or exponentials . for @xmath15
such blobs the deflection angles are summed over all blobs .
therefore the lens can be entirely described by the following parameters : @xmath26 where @xmath27 denotes the parameters for lens blob @xmath13 : @xmath28 and @xmath29 describes the parameters for the sie+@xmath0 component : @xmath30 since the model is not of fixed dimension ( the number of source and lens `` blobs '' is unknown ) , we used a trans - dimensional mcmc sampler based on reversible jump mcmc @xcite .
this framework has been used successfully in many astronomical inference problems ( e.g. * ? ? ?
* ; * ? ? ?
we implemented the mcmc using the rjobject software @xcite , a c++ library for implementing trans - dimensional mcmc with hierarchically - specified priors when the model is a `` mixture model '' or similar , as is the case here .
the rjobject library uses the diffusive nested sampling algorithm ( dns ; * ? ? ?
* ) for its sampling , but the trans - dimensionality is handled by the mcmc moves .
therefore , the marginal likelihood we obtain is one that involves a sum over the hypothesis space for @xmath5 and @xmath31 . in other words
, we do not need separate runs with different trial values of @xmath5 and @xmath31 .
previous astronomical applications of rjobject include @xcite and @xcite .
proposal moves for the parameters ( positions and masses of the source and lens blobs ) , and birth / death moves to add or remove blobs to the source or lens , are handled internally by rjobject and fit into the general framework described in @xcite .
the prior probability distributions for all hyperparameters , parameters , and the data , are given in table [ tab : priors ] , and a factorization of the joint distribution is displayed in figure [ fig : pgm ] . with these priors , we aim to express a large degree of prior uncertainty ( hence the liberal use of uniform , log - uniform , and cauchy distributions ) .
however , we also specify some priors hierarchically to allow ( for example ) blobs to cluster around a typical central position , rather than implying a high probability for substructure positions being spread uniformly ( in a frequency sense ) over the sky .
we also applied hierarchical priors to the substructure masses ( and source fluxes ) so that , while the typical order of magnitude of the masses is unknown , the masses themselves are likely to be roughly the same order of magnitude .
the hyperparameters of these distributions can also be interpreted as straightforward answers to questions about the substructure mass function . [ cols="<,<,<",options="header " , ] we assign somewhat informative ( but heavy - tailed ) cauchy priors to the central positions of the source and the lens .
one might object to the cauchy priors for the central positions ( and argue instead for gaussians ) on the basis that they do not have rotational symmetry .
however , the priors we are assigning are prior only to the values of the data pixel intensities only , and not other facts about the data , such as its dimensions in arc seconds , or its rectangular shape . given these , there is no reason to insist on rotational symmetry .
the cauchy priors are intended to enhance the plausibility ( relative to what a uniform prior would imply ) that the lens and source are somewhere near the centre of the image , but in a cautious way .
the `` circular '' conditional prior for the blob positions has the following density : @xmath32 where @xmath33 .
this results from an exponential distribution ( with expectation @xmath34 ) for the radial coordinate and a uniform distribution for the angular coordinate in plane polar coordinates .
this was chosen ( as opposed to a more `` obvious '' choice such as a normal distribution ) for computational reasons : the rjobject code requires a function to transform from uniform(0 , 1 ) distributions to this distribution and back ( making use of cumulative distribution functions and their inverses ) , and the normal would have required the use of special functions for this . the conditional prior for the data given the parameters is a product of independent gaussian distributions , one for each pixel .
the mean of the gaussian is given by the `` mock '' noise - free image we would expect based on the parameters . the standard deviation of the gaussian is a combination of three terms ; the first ( denoted @xmath35 ) is a `` noise map '' which is loaded from a file , the second is an unknown constant @xmath36 which applies to the whole image , and the third is proportional to the square root of the mock image , with proportionality constant @xmath37 .
this is usually called the `` sampling distribution '' , or sometimes just the `` likelihood '' . however , sampling distribution is misleading since no physical frequency distribution exists which is being sampled from .
the term likelihood is also usually used to refer to the scalar function of the parameters obtained when the data are known . for a discussion of the view that this object is really a prior distribution ,
see @xcite .
it is common to provide a `` variance image '' to lens modelling software , which specifies the standard deviations used in the likelihood function .
however , telescopes do not provide variance images .
it also does not make sense to specify vague priors prior to the image pixel values but posterior to the variance image ( which usually resembles the image itself , and therefore would be highly informative ) .
allowing the standard deviation of the gaussian in the likelihood to depend on the mock image brightness is a more principled way of modelling our actual inferential situation .
nevertheless , we allow for an input variance image as well , which can be used for ad - hoc masking of troublesome regions ( such as unmodelled flux from other sources ) .
computing the posterior distribution over the parameters of such a model requires that we can implement markov chain monte carlo ( mcmc ) over the space of possible sources and lenses . to compute the posterior distribution for the parameters ( of which there are 24464 , depending on the values of @xmath5 and @xmath31 ) , we use diffusive nested sampling @xcite , a form of nested sampling @xcite that uses the metropolis algorithm to move around the parameter space .
the proposal distributions for the blob parameters ( both source and lens ) are handled by the rjobject library @xcite .
this includes birth and death proposals that increase or decrease either @xmath5 , or @xmath31 , as well as proposals that move the blobs ( in their parameter space ) while keeping the number of blobs fixed .
rjobject also facilitates proposals that change the hyperparameters ( either for the lens or the source ) while keeping the actual blobs in place , as well as proposals that change the hyperparameters and shift all of the relevant blobs in so they are appropriate for the new values of the hyperparameters .
evaluating the likelihood function requires that we compute a `` mock '' image from the current setting of the parameters .
this mock image is calculated using standard ray - tracing methods with a uniform grid of @xmath38 rays fired per image pixel .
however , certain kinds of proposals do not affect the image in any way , such as those which change the noise parameters . for efficiency we do not recompute the mock image in these cases .
to demonstrate the method , we generated a simulated dataset where the lens was an sie@xmath39 profile with a single additional substructure .
the image is shown in figure [ fig : image1 ] , and consists of 100 @xmath40 100 pixels .
the point spread function ( psf ) was compact and was defined on a 5 @xmath40 5 grid .
the data was created by firing only one ray per pixel , and the inference was also carried out under the same level of approximation . since the model assumptions are all correct for this image , the demonstration here is purely to illustrate the computational tractability of the problem , and the kind of outputs the method can produce .
the sie@xmath39 parameters were , to three significant figures , @xmath41 , where the length units in the image plane are arbitrary , and rotation angles are measured in radians . the single substructure is located at @xmath42 , about halfway between the center and the upper left image .
its mass is 2.50 units , using the convention where the critical density is 1 . for comparison ,
the sie mass within its critical ellipse is @xmath43 units .
lens close to the center of the image , indicated by the white circle , plus a single substructure close to the top left image ( indicated by a star symbol ) .
the image was blurred by a psf and had some noise added . in an arbitrary system of units
, this image extends from -8 to 8 in the @xmath44 and @xmath45 directions.[fig : image1 ] ] we executed the code to generate 5,000 samples from the `` mixture of constrained priors '' distribution of dns .
our thinning factor was @xmath46 , so @xmath47 mcmc iterations were actually performed , taking approximately 48 hours on a modest desktop pc .
after resampling these samples to reflect the posterior distribution , we were left with 500 ( equally weighted ) posterior samples .
the posterior distributions for complex models , such as the mixture models used here , are often challenging and unintuitive to summarise .
one effective way to visualise the uncertainty in the inferences is to play a movie where each frame is a sample from the posterior distribution .
the degree to which the frames differ from each other conveys the uncertainty remaining after taking the data into account . for the purposes of a paper , static summaries are more convenient than movies .
one useful summary is based on the concept of empirical measure .
the _ empirical measure _ of the substructure positions is a function that takes the actual substructure positions @xmath48 and produces a `` density function '' over two dimensions , composed of delta functions at the positions themselves : @xmath49 intuitively , the empirical measure is a mathematical object that is like an `` infinite resolution histogram '' , in this case a two dimensional histogram , of the substructure positions .
being a function of the actual substructure positions , the empirical measure is not available to us since we do not know those positions with certainty .
however , we have samples from the posterior distribution for those positions , and can use these ( trivially ) to create samples from the posterior distribution for the empirical measure .
we can also summarise this posterior , for example , by taking its expected value .
the posterior expected value of the empirical measure is : @xmath50 where @xmath51 denotes all parameters and hyperparameters ( including the @xmath15s ) and `` @xmath52 '' is an integral and summation over the entire parameter space .
since we can approximate posterior expectations using monte carlo , we can obtain the expected value of the empirical measure using : @xmath53 where @xmath54 is the number of posterior samples .
the resulting function is an `` image '' with a point mass wherever a substructure occurred . for visualisation purposes the image
can be blurred , or calculated at a lower resolution by replacing the dirac - delta function with a discrete version which returns a nonzero constant if a substructure appears in a pixel or zero otherwise .
the masses of the smooth and substructure components of the lens are usually of interest .
since the total mass of an sie lens is infinite , the question needs to be redefined , so we ask about the mass within some aperture of finite area . for an sie ,
the mass ( in dimensionless units based on the critical density ) within the critical ellipse is simply @xmath55 .
however , the blobs have finite total masses @xmath56 . to obtain the posterior distribution for the lens masses
, one must be clear about exactly which mass they are talking about , and many definitions are possible , although some might be more meaningful or well constrained than others .
in the present paper we do not address the question of exactly which quantities related to the density profile of the lens are most scientifically interesting .
nevertheless , we can verify that the results of the inference do behave in understandable ways .
for example , in figure [ fig : masses1 ] , we plot the joint posterior distribution for the sie mass within its critical ellipse and the total substructure mass over the entire domain .
as one would expect , there is a strong dependence between these two quantities in the posterior distribution , as mass in the sie can be traded off with mass in substructures to some extent , while the model remains ( loosely speaking ) `` consistent with the data '' . ]
the posterior distribution for @xmath31 is also clearly of interest , and is displayed in figure [ fig : n_lens1 ] .
the prior for this parameter was uniform from 0 to 10 ( inclusive ) , and the possibility @xmath57 has been ( loosely speaking ) `` ruled out '' by the image data . the true solution ( @xmath58 )
has the highest probability .
however , the possibilities with @xmath59 are still fairly plausible , since it s possible that a very small substructure exists , and it s also possible that two substructures near each other could mimic the effect of one . the degree to which these possibilities are plausible
is related to the choice of prior for the blob amplitudes and positions . ,
the number of substructures in the lens .
the prior was uniform and the true value used to generate the data was 1 .
however , the data is only informative enough to suppress the probability of values above 1 slightly , since it is possible ( given this data ) that low mass substructures might exist somewhere , or that what we think is one substructure might actually be two close together , and other such possibilities .
[ fig : n_lens1 ] ] the estimated marginal likelihood of our model ( averaged over all parameters including @xmath31 and @xmath5 ) is @xmath60 \approx -14141.4 $ ] , and the information ( kullback - leibler divergence from the prior to the posterior ) is @xmath61 nats .
the information represents the degree of compression of the posterior distribution with respect to the prior , and can be interpreted quite literally as how much was learned about the parameters from the data .
it is also straightforward to estimate from nested sampling @xcite .
its definition is : @xmath62 \ , d\boldsymbol{\theta}\end{aligned}\ ] ] and we can build intuition about its meaning based on some simple examples .
one such example is a uniform prior over a volume @xmath63 and a posterior which is uniform over a smaller volume @xmath64 contained within @xmath63 . in this case @xmath65
therefore , a value of @xmath66 nats implies the posterior distribution occupies roughly @xmath67 of the prior volume . ] for the sampled parameter values representing the posterior distribution , there was no structure in the residuals ( not shown ) , when normalised by the noise standard deviation in each pixel .
this is unsurprising since the model assumptions are entirely correct for this dataset .
to further test the algorithm , we created a `` harder '' simulated image , where the lens consisted of an sie+@xmath0 profile plus 10 additional substructures .
the image is shown in figure [ fig : image2 ] . with the easy dataset ,
there was little hope of measuring the substructure mass function parameter @xmath68 since there was only a single substructure which would provide little information about @xmath68 ( only constraining its general order of magnitude ) , even if its mass were measured perfectly .
( whose centre is indicated by a white circle ) and ten substructures .
only five of the substructures ( star symbols ) occured within the square image boundary , but the other five still have an effect on the image .
[ fig : image2 ] ] as with the ` easy ' dataset , we generated 5,000 samples from the dns target distribution and thinned by a factor of @xmath46 , so @xmath47 mcmc iterations were actually performed , taking approximately 60 hours .
the slower runtime in this case was because @xmath69 , more cpu time was spent exploring parts of the hypothesis space where @xmath31 is high .
the run resulted in 557 equally - weighted posterior samples .
the posterior distribution for the sie mass and the total substructure mass is given in figure [ fig : masses2 ] .
the samples can be viewed as a movie at www.youtube.com/watch?v=o3ppfksk248 . , but for the ` harder ' simulated dataset .
[ fig : masses2 ] ] to demonstrate the claim that we can infer something about the mass function of substructures directly from the image data , we have plotted the posterior distribution for @xmath68 , the hyperparameter for the substructure masses , in figure [ fig : mu_lens ] .
the true value was 1 , and the posterior distribution indicates a fairly wide uncertainty range that is at least consistent with the true value in some sense . given that the image only contained a few substructures whose masses could be measured with any accuracy , the large uncertainty is not surprising . , the hyperparameter which determines the prior expected value of substructure masses , given the ` harder ' simulated data .
the true value ( in arbitrary units ) was 1 , which is indicated by the vertical line .
[ fig : mu_lens ] ] , but for the ` harder ' simulated dataset .
the true number of substructures was 10 , which also happens to be the posterior mode .
only five substructures were located within the domain of the image , but @xmath69 is probable because the spread - out positions of the substructures imply that @xmath70 is probably large .
[ fig : n_lens2 ] ] , but for the ` harder ' simulated dataset . for all of the substructures located close to the ring , the positions were well constrained .
there was weak evidence for some substructures that did not in fact exist .
[ fig : substructures2 ] ] the estimated marginal likelihood of our model , for the ` harder ' simulated data , is @xmath60 \approx -5671.9 $ ] , and the information ( kullback - leibler divergence from the prior to the posterior ) is @xmath71 nats . as with the easy dataset , the standardised residuals of the sampled models resembled noise .
as a further demonstration the method , we apply it to the g - band data of the cosmic horseshoe j1004 + 4112 @xcite taken with the isaac newton telescope ( int ) .
the image is shown in figure [ fig : image3 ] .
for this system , more data is available ( i and u band data , as well as more recent hubble space telescope ( hst ) imaging ) , although the g - band image is the highest signal - to - noise of the int images .
the source redshift is @xmath72 , and the lens is a massive luminous red galaxy at @xmath73 .
unfortunately our current implementation does nt allow for multi - band data ( unlike @xcite ) , and is quite slow when running on the larger hst image .
therefore this section should be considered a further demonstration of the technique , and not a thorough study of this system . ]
we generated 5,000 samples from the dns target distribution and thinned by a factor of @xmath46 , so @xmath47 mcmc iterations were actually performed , taking approximately 40 hours .
after resampling , this resulted in 718 equally weighted posterior samples .
although this image is smaller than the simulated data ( @xmath74 pixels ) , we fired @xmath75 rays per pixel for greater accuracy . the joint posterior distribution for the sie mass total substructure mass for the cosmic horseshoe is given in figure [ fig : masses3 ] .
as with the simulated data ( figure [ fig : masses1 ] ) , we see an expected negative correlation between these two quantities . however , the points are not as smoothly distributed in this case ( we discuss this issue further in section [ sec : convergence ] ) .
an upper `` limit '' on the sie mass , with 95% posterior probability , is @xmath76 solar masses .
the algorithm has also found some possibilities where the substructure mass is much greater than this . in these cases ,
the substructures are far from the image it is the environment that is being modelled .
the posterior distribution for @xmath31 ( figure [ fig : n_lens3 ] ) also shows some evidence for more than zero substructures .
in particular , the prior probability for @xmath77 was 10/11 @xmath78 0.91 , whereas the posterior probability is approximately 0.997 .
this corresponds to a bayes factor of about 300 in favour of @xmath77 versus the alternative @xmath79 . , but for the cosmic horseshoe .
the joint posterior distribution for the sie mass ( integrated within its critical ellipse , which is not the critical curve of the lens overall ) , and the total mass in substructures .
the units are defined by the critical density , so that a mass of @xmath80 units would have an einstein radius of one arcsecond . to calculate the masses in solar masses
, we assumed a flat cosmology with @xmath81 , @xmath82 , and @xmath83 km / s / mpc .
[ fig : masses3 ] ] , but for the cosmic horseshoe . the posterior distribution for the number of lens substructures in the cosmic horseshoe system .
there is moderate evidence in favour of the hypothesis that @xmath84.[fig : n_lens3 ] ] despite weak evidence for the existence of substructure , when we examine the expected value of the empirical measure of substructure positions ( figure [ fig : substructures3 ] ) we find no consistency in their positions , unlike for the simulated data ( figure [ fig : substructures1 ] ) .
this situation is not uncommon .
for example , @xcite found evidence for a large number of keplerian signals in a time series , but the number of such signals with well constrained properties was much lower .
related to this , we can compute posterior probabilities for any hypothesis about the substructure masses . with 95% probability
the substructure mass is less than 1.53@xmath85 solar masses and with 75% probability it is less than 5.10@xmath86 solar masses .
as the einstein ring itself implies a mass of about @xmath87 solar masses , these summaries are affected by the possibility of blobs outside the main einstein ring . as with any inference , the results presented here may be sensitive to many of the input modelling assumptions , and a slightly different ( yet still reasonable ) set of choices might yield different results .
for example , if the density profile of the lens was smooth but not of the sie form , and the data were informative enough to show this , the current model would only be able to explain the data by adding substructures .
hence , an alternative explanation for these results is that , rather than containing a substructure , the density profile of the lens is simply not within the sie@xmath39 family .
in fact , some of the posterior samples obtained contain massive substructures outside of the image , which could be modelling higher - order effects of the environment beyond what is captured by the simple external shear model .
this is one way of violating the sie@xmath39 assumption , and another is simply to have a different projected density profile .
one way of further investigating this is to use a different family of smooth lens models and doing model selection based on the marginal likelihoods .
another option which we defer to future work is to implement a more flexible model ( such as a mixture of concentric elliptical lenses ) .
this result is consistent with that of @xcite , who found a slight preference for an elliptical power law profile over the sie ( which is a specific case of an elliptical power law ) .
another potential inadequacy of our model is the assumption that the psf is known . in practice
, the psf is often estimated from the image of nearby star(s ) , or from theoretical knowledge of the telescope ( especially in the case of space based imaging ) . however ,
if the psf is misspecified , or in fact varies across the image , this could induce subtle effects in the imaging which our current model could only explain using substructure . , but for the cosmic horseshoe .
the positions of substructures encountered in the mcmc sampling for the cosmic horseshoe ; technically a monte carlo approximation to the posterior expected value of the empirical measure of substructure positions .
unlike figure [ fig : substructures1 ] , for the cosmic horseshoe there is little consistency in the positions of the substructures .
[ fig : substructures3 ] ] the estimated marginal likelihood of our model is @xmath60 \approx -6948.5 $ ] , and the information ( kullback - leibler divergence from the prior to the posterior ) is @xmath88 nats .
three example models ( lens and source ) sampled from the posterior are shown in figure [ fig : sources ] .
the model was able to fit the data down to the noise level .
the `` effective sample size '' returned by dns , which we have described as the number of posterior samples , takes into account the fact dns s target distribution is not the posterior .
however , it does not take autocorrelation into account , and thus can present an optimistic picture of the accuracy of any monte carlo approximations to posterior quantities .
most standard diagnostic techniques used in mcmc can be applied here .
the simplest of these is a check of reproducibility .
if different runs yield substantially different results , the mcmc output should be treated with caution .
for example , in figure [ fig : masses3 ] , there is a correlation between the mass attributed to the sie component and that attributed to substructures .
however , the distribution also appears somewhat `` lumpy '' or multimodal .
this may not be a feature of the actual posterior distribution , but could arise due to imperfect sampling . whereas a standard metropolis sampler would move around very slowly in the hypothesis space , dns naturally spends a non - negligible fraction of the time sampling the prior .
therefore , a particle exploring the parameter space can `` forget '' its good - fitting position , move somewhere completely different , and find another good - fitting model in a different location .
this is a natural feature of the algorithm ( and is also present in related algorithms such as parallel tempering ) . despite this
, the patchy nature of figure [ fig : masses3 ] suggests that posterior exploration is still challenging in this problem .
the results for the simulated data ( figure [ fig : masses1 ] ) were less complex because of the definite existence of one substructure .
its mass and position provide some information about the hyperparameters @xmath89 , restricting the probability that high mass substructures exist far from the image .
we have developed a trans - dimensional bayesian approach motivated directly by the question of whether dark substructures exist in a lens galaxy , given image data . by making use of the diffusive nested sampling algorithm @xcite and the rjobject library @xcite
, we outsource the difficulties associated with choosing metropolis proposals for a hierarchical model of non - fixed dimension .
the model allows for source and lens `` blobs '' to appear as needed to explain the data .
the prior for the blobs properties is specified hierarchically , to more realistically model sensible prior beliefs , and to tie the model parameters directly to questions of scientific interest such as the mass function of substructures . as a proof of concept
, we demonstrated the successful recovery of a single substructure from simulated data for which all of the model assumptions were true .
we then applied the method to an image of the cosmic horseshoe system and found moderate evidence in favour of the existence of a substructure , or at the very least , a departure from a simple sie+@xmath0 lens profile .
the main output of our method is a set of posterior samples , each representing a plausible scenario for the lens and the source given the data and the assumed prior information .
these samples can be displayed as a movie , which is a very useful method for intuitively understanding the remaining uncertainty .
any posterior summaries of interest can be produced from the samples , but the interpretation of many of these summaries is not necessarily straightforward .
we have suggested that the posterior expectation of the empirical measure of substructure positions is one of the most helpful summaries .
the model assumptions used in this paper are fairly simple . in future work
we intend to generalize the model to multi - band data , and extend the smooth lens model beyond the overly simplistic sie+@xmath0 assumption .
applications to other systems are also forthcoming , as well as variants on the model using different conditional priors for the substructure properties given the hyperparameters .
it is a pleasure to thank phil marshall ( stanford ) , tommaso treu ( ucla ) , ross fadely ( nyu ) , alan heavens ( edinburgh ) , and the referee for valuable discussion .
thomas lumley ( auckland ) also deserves credit for ending bjb s usage of the jet colormap .
this work was funded by a marsden fast start grant from the royal society of new zealand .
|
we introduce a bayesian solution to the problem of inferring the density profile of strong gravitational lenses when the lens galaxy may contain multiple dark or faint substructures .
the source and lens models are based on a superposition of an unknown number of non - negative basis functions ( or `` blobs '' ) whose form was chosen with speed as a primary criterion .
the prior distribution for the blobs properties is specified hierarchically , so the mass function of substructures is a natural output of the method .
we use reversible jump markov chain monte carlo ( mcmc ) within diffusive nested sampling ( dns ) to sample the posterior distribution and evaluate the marginal likelihood of the model , including the summation over the unknown number of blobs in the source and the lens .
we demonstrate the method on two simulated data sets : one with a single substructure , and one with ten .
we also apply the method to the g - band image of the `` cosmic horseshoe '' system , and find evidence for more than zero substructures . however , these have large spatial extent and probably only point to misspecifications in the model ( such as the shape of the smooth lens component or the point spread function ) , which are difficult to guard against in full generality . [ firstpage ] gravitational lensing : strong methods : data analysis methods : statistical
|
launched in june 2008 , the _ fermi _
@xmath1-ray space telescope affords an ideal opportunity to investigate the inner workings of active galactic nuclei ( agn ) . to date , _ fermi _ has spent over 95% of its lifetime in _ all - sky - survey _ observing mode , whereby the large area telescope ( lat ) onboard _ fermi _ points away from the earth and rocks north and south of its orbital plane on consecutive orbits .
this rocking motion of the _ fermi_-lat detector , coupled with it s large effective area , allows _
fermi _ to scan the entire @xmath1-ray sky every two orbits , or approximately every three hours ( @xcite ) .
the scanning ability of the lat has allowed us to catch agn during brief flares of @xmath1-ray activity ( e.g. @xcite ) , with these flares sometimes resulting in the discovery of very high energy ( vhe ; @xmath0 gev ) emission from the agn ( e.g. @xcite & @xcite ) . while its three hour scan period is important for catching brief periods of flare activity from agn , coupling _
fermi_-lat s continual scanning of the sky with a long mission lifetime allows us to construct a deep exposure of the extragalactic @xmath1-ray sky .
this deep exposure affords us the ability to perform searches for faint vhe sources which would otherwise would go undetected by the pointed observations of ground - based imaging atmospheric cherenkov telescope ( iact ) arrays .
utilising such an approach , a recent study of a deep @xmath7 year _ fermi_-lat exposure uncovered vhe emission from the bl lac object rbs 0970 ( @xcite ) .
rbs 0679 , also known as bzb j0543 - 5532 , is a point like radio source , at a redshift of z @xmath8 ( @xcite ) .
detected as a bright x - ray source with the _ rosat _ satellite ( @xcite ) , rbs 0679 has been optically identified as a bl lac object ( @xcite )
. discovered as a source of @xmath1-rays in _
fermi_-lat observations ( @xcite ) , rbs 0679 is included in both the 1 and 2 year lat agn catalogues ( 1lac , @xcite ; 2lac @xcite ) , as well as the recent @xmath9 gev lat catalogue ( @xcite ) .
recently the h.e.s.s .
collaboration has published an upper limit on rbs 0679 of @xmath10 , from 8.1 hours of observations ( @xcite ) .
the energy threshold for this upper limit was 390 gev .
interestingly , the h.e.s.s .
observations found that , even after accounting for absorption of the @xmath1-ray flux by the extragalactic background light ( ebl ) , this upper limit is not compatible with a simple power - law extrapolation of rbs 0679 s 2lac spectrum to above 390 gev .
this suggests the presence of an intrinsic break in the @xmath1-ray spectrum , not associated with the process of ebl absorption . this paper reports the discovery of vhe emission from rbs 0679 . using the same approach that discovered rbs 0970 as a vhe source , 5.3 years of _
fermi_-lat data also revealed three ultraclean @xmath0 gev events clustered within [email protected] of rbs 0679 . unlike rbs 0970
, the vhe emission detected from rbs 0679 does not coincide with severe spectral hardening , with the spectral index of the @xmath5 gev flux during the detection of the vhe photons being consistent with the 5.3 year average .
however , the f@xmath11 parameter space suggests a softer - when - brighter property for the global @xmath1-ray characteristics throughout the 5.3 year data set . in 2
the _ fermi_-lat observations and analysis routines used in this study are described , along with the results of the @xmath12 gev likelihood analysis .
the results of the vhe study of rbs 0679 are shown in 3 .
a brief investigation into the global @xmath1-ray characteristics of rbs 0679 when the vhe emission occurs is presented in 4 , with the conclusions given in 5 .
it should be highlighted that this paper reports the discovery of vhe emission from rbs 0679 .
a follow - up publication with sed modelling and interpretation is currently in preparation and will follow shortly .
the data used in this study comprises all _ fermi_-lat event and spacecraft data taken during the first 5.3 years of _ fermi_-lat operation , from 2008 august 4 to 2013 december 12 , which equates to a mission elapsed time ( met ) interval of 239557417 to 408871812 .
all source @xmath1-ray events - ray ( see @xcite for details on event classification ) . ] , in the @xmath13 gev energy range , within a 13@xmath2 radius of interest ( roi ) centered on the second fermi agn catalogue ( 2lac ; @xcite ) position of rbs 0679 , ( @xmath14 , @[email protected] , @[email protected] ) , were considered . in accordance with the pass7_rep criteria , a zenith cut of 100@xmath2 was applied to the data to remove any cosmic ray induced @xmath1-rays from the earth s atmosphere .
the good time intervals were generated by applying a filter expression of `` ( data_qual@xmath171 ) & & ( lat_config@xmath171 ) & & abs(rock_angle)@xmath18 52 '' to the data , where the ( data_qual ) and ( lat_config ) flags remove sub - optimal data affected by spacecraft events and the ( abs(rock_angle ) ) flag removes data periods where the lat detector rocking is greater than 52@xmath2 . throughout this analysis , version v9r32p5 of the _ fermi science tools _ was used in conjunction with the p7rep_clean_v15 and p7rep_source_v15 instrument response functions ( irfs ) .
the irfs defines how the lat detector performs for various parameters such as incident photon energy , incident angle and location of photon conversion within the detector . during the analysis , a model file consisting of both point and diffuse sources of @xmath1-rays
was employed .
in particular , the model file consisted of the most recent galactic and extragalactic diffuse models , and all @xmath1-ray point sources within a 18@xmath2 roi centered on rbs 0679 .
the positions of these point sources , along with their spectral shapes , were taken from the second fermi source catalogue ( 2fgl ; @xcite ) . located [email protected] from rbs 0679 ,
the extended @xmath1-ray emission associated with the large magellanic cloud was accounted for with the _ fermi_-lat collaboration s lmc.fits spatial model , with the normalisation and spectral parameters frozen to their 2fgl values .
the normalisation factor of the extragalactic diffuse emission was left free to vary , while the galactic diffuse template was multiplied by a power law in energy , the normalisation of which was left free to vary .
firstly a binned maximum likelihood analysis was performed on the entire 5.3 year data set in the @xmath12 gev energy range . for rbs 0679 itself ,
a power law spectral shape of the form @xmath19 a@xmath20 was assumed , with the normalisation , a , and the spectral index , @xmath21 , left free to vary .
the normalisation and spectral parameters of all point sources within 13@xmath2 of rbs 0679 were left free to vary , while the normalisation and spectral parameters for all point sources within an annulus of 13@xmath2 to 18@xmath2 from rbs 0679 were fixed to those published in the 2fgl catalogue . utilising the above described model
, the binned likelihood analysis of the 5.3 year data set , utilising the p7rep_source_v15 irf , resulted in the following best - fit power law function for rbs 0679 : @xmath22 @xmath23 which equates to an integrated flux , in the @xmath12 gev energy range , of @xmath24 taking only statistical errors into account .
for the best - fit power law description , a test statistic$ ] , where @xmath25 and @xmath26 are defined as the likelihood when the source is included or not respectively ( @xcite ) . ] of @xmath27 was found , corresponding to a @xmath28 detection of rbs 0679 in the @xmath12 gev energy range over the 5.3 year period .
the long exposure of the analysed observations can result in additional faint sources being present in the data that are not present in the 2fgl .
if these sources are present in the data , and not properly accounted for within the model file utilised during the likelihood analysis , they can artifically increase the significance of the @xmath1-ray flux from rbs 0679 ( e.g. @xcite , @xcite & @xcite ) . to check
if indeed any additional sources were present , a residual count map was constructed in the @xmath12 gev energy range . shown in figure 1 ,
the residuals map was obtained by subtracting a ` model ' map gev likelihood fit of equation 1 .
] from the observed counts map , and then dividing by the model map again .
the residuals map shows fluctuations in the count residuals at the @xmath29% level .
these fluctuations are consistent with random fluctuations and as such , the model used during the analysis is an accurate representation of the observed @xmath1-rays , with no new @xmath1-ray point sources being present .
model)@xmath30model ) , in the @xmath12 gev energy range , during the entire 5.3 year period , centered on the co - ordinates of rbs 0679 .
the colour scale is in units of percentage , with fluctuations in the residuals between the model and observed counts map being at the @xmath29% level .
these fluctuations are consistent with random fluctuations and as such , show the model used during the analysis to be a good representation of the data .
it is worth highlighting that the biggest percentage difference between the model and observed count maps , seen at the top of the residuals map , is positionally coincident with the 2lac source pmn j0531 - 4827.,width=264 ]
[ cols="^,^,^,^,^,^ " , ] [ indexdetails ]
with 5.3 years of _ fermi_-lat data , rbs 0679 has been found to be a source of vhe @xmath1-ray photons . with three ultraclean @xmath0 gev photon events within 0.1@xmath2 of rbs 0679 , an unbinned likelihood analysis revealed the significance of this discovery to be at the @xmath31 confidence level . the 5.3 year
integrated @xmath0 gev flux was found to be @xmath32 .
an investigation of the global @xmath5 gev @xmath1-ray characteristics during the 5.3 years revealed a ` softer - when - brighter ' trend .
interestingly , the 28-day periods during which the vhe photons were detected had a spectral index that was consistent with the 5.3 year average , suggesting that the observed vhe emission is not associated with a spectral hardening event .
the detection of three ultraclean events within 0.1@xmath2 of rbs 0679 , coupled with the results of the @xmath6 gev unbinned likelihood analysis , suggest the vhe detection of rbs 0679 is a robust result . while the extrapolation of rbs 0679 s spectrum , as reported in previous _ fermi_-lat catalogues , is inconsistent with the recently announced h.e.s.s .
upper limit , the extrapolation of the best - fit power law for the vhe discovered here is consistent with the h.e.s.s . upper limit .
however , given the large uncertainty on the spectral index of this power - law fit does not significantly rule out the presence of a cut - off in the @xmath1-ray spectrum . to investigate the presence of a cut - off in detail requires follow - up observations with iact arrays such as hess - ii or the future cta ( @xcite ) , in conjunction with indepth sed modelling .
this indepth sed modelling is currently ongoing , the results of which we defer to a later publication .
we thank the referee for their comments and suggestions that has improved the quality of this paper .
this work was undertaken with the financial support of durham university and has made use of public _ fermi _ data obtained from the high energy astrophysics science archive research center ( heasarc ) , provided by nasa s goddard space flight center .
this work has also made use of the nasa / ipac extragalactic database ( ned ) , which is operated by the jet propulsion laboratory , caltech , under contract with the national aeronautics and space administration .
we thank the _
fermi_-lat collaboration for the quality of the data and analysis tools that were used in this study .
99 abdo , a. , et al .
2009 , apj , 707 , 1310 abdo , a. a. , et al . 2010a , apjs , 188 , 405 abdo , a. a. , et al .
2010b , apj , 715 , 429 abramowski , a. , et al .
2010 , a&a , 520 , 83 abramowski , a. , et al .
2013 , a&a , 559 , 136 abramowski , a. , et al .
2014 , a&a , 562l , 4h acharya , b.s . ,
2013 , aph , 43 , 3 ackermann , m. , et al .
2011 , apj , 743 , 171 ackermann , m. , et al .
2012 , apjs , 201 , 4 ackermann , m. , et al .
2012 , apjs , 203 , 4 ackermann , m. , et al .
2013 , apjs , 209 , 13 aliu , e. , et al .
2012 , apj , 750 , 94 anderson , m. w. b. & filipovic , m. d. 2009 , serbian astronomical journal , 179 , 7 atwood , w.b . , et al .
2009 , apj , 697 , 1071 brown , a.m. , 2006 , phd thesis , univ .
durham brown , a.m. , 2013 , mnras , 431 , 824 brown , a.m. , 2014 , mnras , 442 , l56 brown , a.m. & adams , j. , 2011 , mnras , 413 , 2785 brown , a.m. & adams , j. , 2012 , mnras , 421 , 2303 coppi , p.s .
& aharonian , f. , 1998 , 19th texas symposium of relativistic astrophysics and cosmology , ed .
j. paul , t. montemerle & e. aubourg dickinson , h. & farnier , c. , 2013 , a&a , 552 , 134 franceschini , a. , rodighiero , g. & vaccari , m. , 2008 , a&a , 487 , 837 kataoka , j. , et al .
2010 , apj , 715 , 554 krawczynski , h. et al .
2004 , apj , 601 , 151 macias - ramirez , o. , gordon , c , brown , a.m. & adams , j. 2012 , phrvd , 86 , 6004 mattox j.r .
, et al . 1996 , apj , 461 , 396 nandikortjur , g. , jahoda , k.m . , hartman , r.c . ,
mukherjee , r. , sreekumar , r. , bottcher , m. , sambruna , r.m .
& swank , j.h .
2007 , apj , 567 , 706 neronov , a. , semikoz , d. , & vovk , ie . , 2010 ,
a&a , 519 , 6 neronov , a. & semikoz , d. , 2010 , phys .
d , 80 , 123012 nolan , p.l . , et al . 2012 , apjs , 199 , 31 ong , r. & fortin , p. , 210
, atel 2272 pita , s. , et al .
2014 , a&a , 565 , 12 tanaka , y. t , et al .
2013 , apj , 777 , 18 voges , w. , et al . 1999 ,
a&a , 349 , 389
|
in this paper we report the _ fermi _ large area telescope ( lat ) detection of very high energy ( vhe ; @xmath0 gev ) @xmath1-ray emission from the bl lac object rbs 0679 .
5.3 years of lat observations revealed the presence of three vhe photon events within [email protected] of rbs 0679 , with a subsequent unbinned likelihood analysis finding rbs 0679 to be a source of vhe photons at @xmath3 standard deviations ( @xmath4 ) .
an unbinned likelihood analysis of the @xmath5 gev data , binned in 28-day periods , finds both the flux and spectral index to be variable , with a ` softer - when - brighter ' trend in the global @xmath1-ray characteristics . on the other hand ,
the 28-day periods in which the vhe photons were detected have spectral indices that are consistent with the 5.3 year average suggesting that the observed vhe emission is not associated with a spectral hardening event .
the discovery of rbs 0679 in the @xmath6 gev energy range , combined with the non - detection above 390 gev with the h.e.s.s . telescope array , suggest rbs 0679 to be an intriguing source that requires further follow - up observations with ground - based @xmath1-ray observatories .
[ firstpage ] radiation : non - thermal galaxies : active bl lacertae individual ( rbs 0679 ) galaxies : jets gamma rays : galaxies .
|
40004000 = 1000 # 10=to 0.50#1msf= msf=1000 0.35 in the nuclear impact on cosmology : the diagram 0.2 in craig j. copi@xmath13 and david n. schramm@xmath14 0.2 in _
@xmath15department of physics + the university of chicago , chicago , il 60637 - 1433 _ + 0.1 in _ @xmath16nasa / fermilab astrophysics center + fermi national accelerator laboratory , batavia , il 60510 - 0500 _ + 0.1 in _
@xmath17department of astronomy & astrophysics + enrico fermi institute , the university of chicago , chicago , il 60637 - 1433 _ +
in 1974 , gott , gunn , schramm , and tinsley ( hereafter , ggst ) showed that a plot of the hubble constant , @xmath3 , verse the the dimensionless density parameter , , [ eqn : omega ] where @xmath18 is the mass density and = is the critical cosmological density , well illustrated the issues in physical cosmology , particularly for models with cosmological constant @xmath19 .
twenty years later we again use the diagram and show that the constraints of ggst have not changed significantly but the interpretation now illustrates the critical issues in physical cosmology today , namely the dark matter and age problems . as nuclear / particle astrophysicists we note with pride ( or fear ) how many of the most significant lines on the diagram have their origin in nuclear physics arguments .
it will be shown how the diagram dramatically illustrates that there are at least two dark matter problems , namely the bulk of the baryons are dark and the bulk of the matter in the universe is non - baryonic .
it will also be shown that these two dark matter problems persist regardless of the value of @xmath3 .
these arguments center on the big bang nucleosynthesis ( bbn ) constraints on @xmath0 in baryons , .
we will also review the age constraints from globular cluster ages and from nucleocosmochronology .
we will show with the @xmath20 diagram that @xmath21 and the cosmological constant @xmath6 requires @xmath22 .
variations in age - relationships for non - zero @xmath8 are also discussed .
a lower bound on @xmath23 from carbon - detonation powered type ia supernova is also plotted . for comparison ,
the information on the baryon content of hot x - ray emitting gas in clusters based on rosat measurements is also discussed .
this paper will now go through each of the constraints in turn and generate appropriate diagrams .
the friedmann - robertson - walker ( frw ) cosmological model provides a simple physical and mathematical model for describing the large scale structure of the universe by assuming the universe is isotropic and homogeneous .
the smoothness of the background radiation is striking confirmation that the universe is isotropic at a level of 1 part in @xmath24 ( cobe ) .
the assumption of homogeneity is less straight forward to confirm , however , measurements of peculiar velocities of galaxies on very large scales as well as radio source count studies seem to indicate that it is valid . within the frw model
the distance and time scales can be related to the hubble constant , h_0 = , where @xmath25 is the scale factor of the universe and @xmath26 is the present age .
as we shall see the value of the hubble constant is still quite uncertain .
thus in practice it is useful to introduce a dimensionless factor h = to express this uncertainty .
the geometry of the universe is encoded in @xmath0 ( and @xmath8 ) ; for @xmath27 ( @xmath28 ) and @xmath19 the universe is opened and hyperbolic ( closed and spherical ) and for @xmath29 it is opened and flat . for the most part we shall assume the vacuum has no density nor pressure , in other words we assume @xmath30 unless explicitly stated to the contrary . in the frw ,
@xmath19 model the present age of the universe is t_0 = , where f(_0 ) = \ { ll , & _ 0 > 1 + 23 , & _ 0 = 1 + , & _ 0 < 1 . , and @xmath31 is the present value of @xmath0 .
the most stringent limits on the age of the universe come from the age of globular clusters and nucleocosmochronology . the age of globular clusters is essentially determined by the rate of hydrogen burning in low mass , metal poor stars .
when the core of the star has been completely converted to helium the star changes its structure and no longer lies on the main sequence of a hertzsprung - russel luminosity - temperature plot .
the main sequence turnoff age is dependent on numerous aspects of the globular clusters , such as metallicity , helium diffusion in the stars , and the initial helium abundance .
while many groups have calculated globular cluster ages in the @xmath32 range , shi has shown that reasonable systematic assumptions in the calculations could lower the ages to @xmath33 with a firm lower bound of @xmath34 .
this is also consistent with an independent study by chaboyer .
this lower bound can be obtained trivially by noting that globular clusters should be burning hydrogen less rapidly than the sun since they have lower metallicity and a slightly lower mass .
our sun will exhaust the hydrogen in its core in about @xmath35 ( based on the hydrogen burning rate verified by the calibrated gallex experiment@xcite ) .
thus globular clusters must have a lower bound on their age of @xmath34 as quoted above .
nucleocosmochronology provides information about time scales over which the elements in the solar system were formed .
this method couples knowledge of present day abundance ratios , production ratios , and lifetimes of long lived radioactive nuclides .
the standard methods of determining nucloechronometric ages rely heavily on the adopted galactic evolution model .
this can lead to large errors in the deduced galactic age .
an alternative approach is to employ less restrictive , model - independent nucleocosmochronology which studies the constraints that can be made without specific reference to galactic nucleosynthesis models .
when using radioactive decay alone only a strict lower bound is possible .
in particular the mean age of the longest lived chronometer , is longer lived in its ground state , its lifetime is dependent on its thermal environment
so is not useful for a lower bound .
it does , however , constrain the maximum mean age . ]
relative to , can be used to give an extremely conservative lower bound@xcite of @xmath36 . since this
bound assumes all nucleosynthesis occurs in a single event , it is obviously too extreme .
we know that , , and all existed in measurable abundances when the solar system formed 4.6 ago and free decay from a single production event several gyr earlier would not be consistent ( for example , has a half - life of only 17 and only 82 ) .
thus we know the production was more spread out than a single event .
meyer and schramm quantified this spreading out to show that the lower bound from chronology was @xmath37 and subsequent analysis by schramm using improved limits on the production ratios pushed the bound up to @xmath38 .
the results from globular cluster and nucleocosmochronology provide a consistent lower bound for the age of the universe .
we note that globular cluster ages and nucleocosmochronology do not provide a strong upper bound to the age of the universe since one could in principle add several gyr of formation time to any such age determination
. the lower bound does not have these problems since the extreme limit is globular cluster formation on a kelvin - helmholtz collapse time scale , @xmath39 at recombination , @xmath40 which yields formation times @xmath41 after the big bang .
it is not possible to form globular clusters earlier than this time .
based on the above results the age of the universe is constrained to be t_010 .
the resulting excluded region in the plane is shown in figure [ fig : main - fig ] for @xmath19 , figure [ fig : lambda0.4 ] for @xmath42 , and figure [ fig : lambda0.8 ] for @xmath43 . here
@xmath44 is defined as above ( [ eqn : omega ] ) with @xmath45 .
standard homogeneous bbn accurately predicts the primordial abundances of the light elements over 9 orders of magnitude in terms of a single parameter , the density of baryons , @xmath46 . for the constraints on we adopt the recent determination by copi , schramm , and turner , 0.01 h^2 0.02 .
[ eqn : bbn - constraint ] note that @xmath46 is independent of the hubble constant , thus the hubble constant enters into @xmath47 only through .
the curves defined by this choice and the excluded regions are shown in figure [ fig : main - fig ] .
attempts to by - pass these constraints with inhomogeneous models have been shown to fail in that the constraints on the light element abundances yield essentially the same constraints on @xmath47 .
recently tytler and fann have observed deuterium in a quasar absorption system that , if confirmed , restricts the baryon density to an extremely tight range near our quoted upper limit .
it might be noted that one significant difference between our diagram and that of ggst is that in 1974 bbn only provided an upper bound to from deuterium , whereas now we also have a lower bound on from 3 plus deuterium arguments and we have the strict lithium constraints adding consistency to the picture .
the most straight forward method of estimating @xmath0 is to measure the luminosity of stars in galaxies and estimate the mass to light ratio , @xmath48 .
the mass density of visible objects is given by = , where is the average luminosity density in the wavelength band used to determine @xmath48 .
for our galaxy = ( 1.00.3 ) h 8 l_^-3 ml = ( 6 3 ) . here
@xmath49 is the mass of the sun and @xmath50 is the luminosity of the sun .
these values give 0.002 h 0.006 .
this value agrees well with the determination of @xmath48 using the visible part of external galaxies .
the curves defined by this range and the excluded regions are shown in figure [ fig : main - fig ] .
numerous methods for dynamically measuring the density of the universe have been developed all of which give complimentary results .
a review of many of these methods can be found in peebles .
we shall highlight a few of these methods .
the simplest means of measuring mass inside a radius , @xmath51 , is via kepler s third law gm(r ) v^2 r. if the mass were solely associated with the light then we would expect @xmath52 for some object outside of the core of the galaxy . instead
it is observed that @xmath53 for objects far from the center of the galaxy .
this indicates that dark matter exists in a halo around the galaxy .
typical estimates of the mass in halos from this method gives @xmath54 .
this dark - matter could in principle be dark baryons ; however , see the discussion on machos below . note that estimates of the mass density from dynamics scale with @xmath55
as does , thus @xmath0 is independent of @xmath3 .
measurements of average velocity dispersion and average separation of galaxies in clusters provide a means of assessing the amount of matter associated with clusters .
it is generally observed that the velocity of galaxies approach a constant value for large distances from the cluster core . as noted above this
indicates the presence of significantly more mass than is visible in the galaxies themselves . a detailed statistical analysis of galaxy dynamics yields = 0.150.06 .
an independent method of verifying this value of @xmath56 is due to the observation of giant luminous arcs by lynds and petrosian .
these arcs are the image of a bright background object that falls in the line of sight of a cluster core .
the mass of the cluster serves as a gravitational lense of this background object producing the arc .
although the modeling of the mass distribution in the cluster can be quite complex , the general prediction of @xmath57 is in good agreement with the above value .
finally we note that many of these methods can be applied on even larger scales .
for example , the peculiar velocities of clusters of galaxies can be studied similar to what has been done for galaxies .
the result of these types of studies is a consistent bound of @xmath58 . to be conservative
we shall adopt > 0.1 .
this limit is shown in figure [ fig : main - fig ] .
in addition to providing a measure of @xmath0 , clusters also provide a means of estimating .
the three main mass components of clusters of galaxies are star in galaxies , hot intracluster gas , and dark matter .
the first two are comprised solely of baryons .
optical observations provide an estimate of the baryon mass in stars and x - ray studies provide an estimate of the baryons in hot gas .
these two quantities together provide an estimate to the total baryon mass in the cluster .
the total mass of the cluster is more difficult to determine .
it is sensitive to numerous assumptions .
in particular the cluster is typically assumed to be spherical and in dynamical equilibrium ( virialized ) . if either of these assumptions are not valid the derived total mass could be incorrect .
frequently structure formation models are employed to remove some of this sensitivity .
white , navarro , evrard , and frenk employed a `` standard '' cold dark matter ( cdm ) model ( @xmath21 ) coupled with optical and x - ray studies to deduce a baryon fraction for the coma cluster of = ( 0.0090.002 ) + ( 0.05 0.01 ) h^-3/2 . here
the first term is due to baryons in stars and the second to baryons in hot gas .
note that @xmath59 is defined by = , where @xmath60 is the mass in baryons of the cluster and @xmath61 is its total mass .
the region defined by these limits is shown in figure [ fig : omegabc ] .
if clusters are a fair sample of the universe then we expect @xmath62 which is clearly not the case if @xmath63 .
the question of whether galaxies in clusters trace the dark matter is still an open one .
babul and katz found that baryons in an @xmath21 cdm model are more strongly concentrated than the dark matter .
thus @xmath64 and there is no inconsistency in the results .
alternate models with some admixture of hot dark matter also yield @xmath65 . at present
there are still a number of difficulties to be worked out in the interpretation of the x - ray gas in clusters result .
it is clear , however , that this observation provides important constraints on cluster formation models .
a well known feature of frw cosmologies is at an epoch @xmath66 if @xmath67 ( @xmath68 ) the universe evolves towards @xmath69 ( @xmath70 ) on a time scale @xmath71 ( see ref . ) .
notice that @xmath72 is an unstable equilibrium point .
at early times @xmath25 was changing rapidly and @xmath73 was large .
thus all evolutionary changes occurred on much shorter time scales . since the universe is clearly not more than an order of magnitude away from @xmath72 today it must have been unity to high accuracy in all earlier epochs .
in particular we have a good understanding of the physics of the universe at the beginning of bbn ( @xmath74 ) .
if we trace @xmath0 to the epoch of bbn we find that @xmath0 must have been unity to @xmath75 decimal places at that time .
the extreme amount of tuning required to satisfy this is an initial condition within the standard big bang model .
the theory of inflation succinctly explains this fine tuning , the homogeneity and isotropy of the universe , and other initial conditions with physics motivated by particle physics theory .
most models of inflation require the universe to be perfectly flat , @xmath76 , or at least @xmath77 .
though this is not a measurement and is an untested theory it provides a compelling theoretical argument for @xmath21 .
furthermore , recent measurements on the largest scales indicate that @xmath78 .
though there is no firm evidence , we show our bias towards @xmath21 by plotting this value as a dashed line in figure [ fig : main - fig ] and [ fig : omegabc ] and the line @xmath79 in figures [ fig : lambda0.4 ] and [ fig : lambda0.8 ] .
the observational determination of @xmath3 has a long and interesting history beginning with the original measurement by hubble@xcite of @xmath80 . since this time
the value has been reduced by about a factor of @xmath81 due to a number of systematic errors in the assumptions used by hubble .
currently the measurements fall into two distinct groups : @xmath82 ( see ref . ) and @xmath83 ( see ref . ) .
these two values are represented by dashed lines in figure [ fig : main - fig ] . at present possible systematic errors
do allow for a consistent resolution of the two values at @xmath84 which happens to also be near the value recently obtained by riess , press , and kirshner .
notice from figure [ fig : lambda0.4 ] that @xmath85 can be made consistent with a flat universe if @xmath86 .
the main difficulty in determining @xmath3 is establishing the distance to an object .
although it is relatively easy to establish the redshift of an object , its absolute distance is quite difficult to determine .
the redshifts for objects observed by hubble are the same today whereas the distances hubble estimated are quite far from present estimates .
the traditional approach for absolute distance measurements is to identify a standard candle ( an object with a known luminosity ) and use its apparent luminosity to determine a distance . in practice a given standard candle
can only be observed over a limited distance range .
thus a ladder of distances to known objects must be built starting with nearby objects .
each rung on the distance ladder is governed by a different standard candle .
slight errors or disagreements on an early rung can correspond to large uncertainties and differences for very distant objects and thus different values for the hubble constant .
one method of minimizing this problem is to use type ia supernovae where the luminosities are known , at least roughly , from the physics and they can be observed over a large range of distances .
detailed calibrations of type ia supernovae involves establishing distances by other techniques , including cepheid variable stars from the hubble space telescope , to nearby galaxies where such supernovae have exploded .
this method tends to give a value of @xmath87 .
note that this technique can also be used to bound @xmath3 from below .
the extreme lower bound comes from the fact that type ia supernovae seem to be caused by a c - o white dwarf star burning its c - o to fe via carbon detonation / deflagration . assuming that the entire @xmath88 chandrasekhar mass of the white dwarf is pure carbon and is completely converted to iron provides the maximal energy release and the limit h_0 38 .
we show this lower limit in figure [ fig : main - fig ] .
the so called two dark - matter problems are ( i ) most of the baryons in the universe are dark ( baryonic dark matter ) and ( ii ) most of the matter in the universe is non - baryonic ( non - baryonic dark matter ) . both of the problems are illustrated in figure [ fig : main - fig ] .
note that regardless of the value of @xmath3 , does not intercept .
similarly on the high @xmath0 side , only marginally intersects at very low @xmath3 values .
thus except for @xmath89 we already know that requires non - baryonic dark matter .
high values of @xmath3 amplify the need for non - baryonic dark matter . from figure
[ fig : main - fig ] we see that the regions defined by and do not overlap for any value of @xmath3 shown
. thus even if all of the visible matter is baryonic , most ( at least 70% ) of all baryons must be dark .
although the halo of our galaxy could be composed of dark baryonic objects such as brown dwarfs and jupiters , known as massive compact halo objects ( machos ) , gates , gyuk , and turner using the relative paucity of microlensing events in the direction of the large magellanic cloud and the high number of such events towards the galactic bulge have argued that less than 40% of the halo can be machos indicating that the halo probably also includes a non - baryonic component .
this latter point seems to require that at least some of the non - baryonic dark matter must be cold so it can condense in halos .
one loop hole to this is black holes with @xmath90 , too small to tidally disrupt star clusters and yet big enough to avoid over production of heavy elements . as noted above , if @xmath91 the measurement of already requires non - baryonic dark matter .
moreover , the observational evidence suggesting @xmath78 and the theoretical arguments of avoiding fine tuning ( such as from inflation ) requiring a flat universe further strengthens this requirement . in the case of @xmath78
non - baryonic dark matter is required for all values of @xmath3 since @xmath92 for any value of @xmath3 .
notice that even for @xmath93 ( figure [ fig : lambda0.8 ] ) the baryonic and non - baryonic dark matter arguments are unchanged .
we have used the diagram to illustrate the nature of key problems in physical cosmology and the role nuclear physics plays in them .
we have seen how bbn serves to define the two dark matter problems .
we have seen that if @xmath94 then @xmath95 requiring the existence of non - baryonic dark matter . if @xmath63 as current observational and theoretical work indicates then non - baryonic dark matter is required for all values of @xmath3 .
finally we have reviewed the age constraints from globular clusters and nucleocosmochronology to show that @xmath96 requires @xmath97
we would like to acknowledge our previous collaborators d. n. dearborn , j. r. gott , j. e. gunn , b. s. meyer , k. a. olive , p. j. e. peebles , g. steigman , m. s. turner , x. shi , j. w. truran , and g. j. wasserburg whose work we have drawn upon .
we also gratefully remember beatrice tinsley who was the person who brought the ggst collaboration together . this work has been supported in part by nsf grant ast 90 - 22629 , doe grant de - fg02 - 91-er40606 , and nasa grant nagw-1321 and by a nasa gsrp fellowship .
the plane showing allowed and excluded regions .
shown here are the limits based on the age of the universe , @xmath26 , the fraction of visible matter in the universe , , the fraction of baryons in the universe , , the fraction of matter in clusters of galaxies , , and type ia supernovae , @xmath23 . also shown as dotted lines are the two current values for the hubble constant , @xmath98 and @xmath99 , and the theoretically preferred @xmath21 . in this figure @xmath19
|
the diagram is resurrected to dramatically illustrate the nature of the key problems in physical cosmology today and the role that nuclear physics plays in many of them . in particular it is noted that the constraints on from big bang nucleosynthesis do not overlap with the constraints on nor have significant overlap with the lower bound on @xmath0 from cluster studies .
the former implies that the bulk of the baryons are dark and the later is the principle argument for non - baryonic dark matter .
a comparison with hot x - ray emitting gas in clusters is also made .
the lower bound on the age of the universe from globular cluster ages ( hydrogen burning in low mass stars ) and from nucleocosmochronology also illustrates the hubble constant requirement @xmath1 for @xmath2 .
it is also noted that high values of @xmath3 ( @xmath4 ) even more strongly require the presence of non - baryonic dark matter
. the lower limit on @xmath3 ( @xmath5 ) from carbon detonation driven type ia supernova constrains long ages and only marginally allows to overlap with .
diagrams of for @xmath6 and @xmath7 are presented to show that the need for non - baryonic dark matter is independent of @xmath8 .
= = 6.5 in = -0.5 in # 1 [ # 1 ] # 110^#1 @xmath9 # 1@xmath10 # 1#1@xmath11 # 1#2 # 1#2 # 1 # 1#1 .
+ # 1#23.6pt @bibitem = bibitem bibitem#1#2ifundefinedr@#1warning multiple entries for reference ` # 1'ifundefinedb@#1warning reference ` # 1 ' not citednamedefr@#1#2 @lbibitem = lbibitem lbibitem[#1]#2#3ifundefinedr@#2warning multiple entries for reference ` # 2'ifundefinedb@#2warning reference ` # 2 ' not citednamedefr@#2#3
namedefs@#2#1 putbibrefs#1makearef # 1 * , makearef#1,#1 = = makearef ifundefineds@#1 ifundefinedr@#1citation label ` # 1'?nameuser@#1 citex[#1]#2@fileswauxout citeaciteforciteb:=#2#1 # 1ifundefinedd@#1addlabel#1warning label ` # 1 ' cited . not necessary .
addlabel#1ifundefined@bibrefslistbibrefslist#1 , bibrefslistbibrefslist#1 , @cite = cite @cite#1#2@xmath12 @cite#1#2#1@tempswa,#2 # 1cite=@citemsfmsf = 1000 # 1
|
a detailed study of the beam energy ( @xmath4 ) , transverse momentum ( @xmath0 ) and system size dependence of identified hadron production will provide the necessary data to understand the mechanism of energy loss , and put constraints on parameters in energy loss model calculations like initial gluon density @xcite and life time of dense matter @xcite .
they will also help in probing the difference in the energy loss of quarks and gluons in the medium @xcite .
the energy dependence of baryon to meson ratio at the intermediate @xmath0 region from 2 - 6 gev/@xmath6 will address the specific prediction from the quark coalescence models of a higher baryon to meson ratio at @xmath4 = 62.4 gev compared to @xmath4 = 200 gev @xcite . at high @xmath0 ( @xmath7 6 gev/@xmath6 )
the @xmath2/@xmath8 ratio can provide information on quark and gluon jet conversions in the medium formed in heavy ion collisions @xcite .
the data presented here are obtained from the time projection chamber ( tpc ) and the time - of - flight ( tof ) detector in the star experiment @xcite at rhic in the year 2004 .
measurements of the ionization energy loss of charged tracks in the tpc gas under a magnetic field of 0.5 tesla are used to identify @xmath1 , @xmath2 and @xmath3 within @xmath9 @xmath5 0.5 and full azimuth , for @xmath10 gev/@xmath6 and @xmath11 gev/@xmath6 .
the data from tof provides particle identification up to @xmath12 gev/@xmath6 for pions and protons in @xmath13 in pseudorapidity and @xmath14 rad in azimuth . weak - decay feed - down ( e.g. @xmath15 ) to the pion spectra
was subtracted from the results presented .
inclusive @xmath2 and @xmath3 production include all of the hyperon feed - downs to reflect the total baryon production .
systematic errors for the tof measurements are similar at both energies and are around 8% @xcite .
the total systematic errors on charged pion yields at both energies and collision systems are estimated to be @xmath16 15% and those for proton and anti - protons to be @xmath16 26% over the entire @xmath0 range studied @xcite .
one nuclear modification factor ( @xmath17 ) is defined as @xmath18 , where @xmath19 is the average number of binary nucleon - nucleon collisions per event .
figure [ fig1 ] shows the beam energy and @xmath0 dependence of @xmath17 for @xmath20+@xmath21 and @xmath2+@xmath3 in au+au collisions . there is a distinct difference in the @xmath0 dependence of @xmath17 for charged pions and protons+anti - protons observed at both energies .
the @xmath17 values are higher at @xmath4 = 62.4 gev than for @xmath4 = 200 gev for @xmath0 @xmath5 7 gev / c ; beyond this @xmath0 they seem to approach each other . considering a lower parton energy loss due to a smaller initial gluon density at the lower energy
, one may expect the @xmath17 to approach each other at high @xmath0 due to a steeper initial jet spectra at 62.4 gev . at high @xmath0 a similar value of @xmath17 for @xmath2+@xmath3 and @xmath22
is observed at both energies .
this may indicate a similar energy loss of quarks and gluons in the medium formed in high energy heavy ion collisions . for @xmath20+@xmath21 and @xmath2+@xmath3 in au+au collisions at @xmath4 = 200 ( left panel ) and 62.4 gev ( right panel ) .
the error bars are statistical , the shaded bands are systematic errors .
there is an additional error of about 10% not shown in the figure due to uncertainty in @xmath23 calculations.,title="fig : " ] for @xmath20+@xmath21 and @xmath2+@xmath3 in au+au collisions at @xmath4 = 200 ( left panel ) and 62.4 gev ( right panel ) .
the error bars are statistical , the shaded bands are systematic errors .
there is an additional error of about 10% not shown in the figure due to uncertainty in @xmath23 calculations.,title="fig : " ] figure [ fig2 ] shows the @xmath24 and @xmath25 ratios for au+au collisions at @xmath4 = 200 and 62.4 gev .
the experimental results are compared to model calculations based on coalescence and jet fragmentation @xcite .
the @xmath24 ratio for au+au collisions at @xmath4 = 62.4 gev is higher than the corresponding values at @xmath4 = 200 gev in the intermediate @xmath0 range .
the case for @xmath3/@xmath21 ratio is reversed .
the higher value of @xmath24 at intermediate @xmath0 in 62.4 gev compared to 200 gev is consistent with quark coalescence models @xcite .
however a detailed comparison of the @xmath24 and @xmath25 ratios to predictions from model shows there is a lack of quantitative agreement .
a shift in the peak position for the ratios at 62.4 gev predicted by the models are not observed in the data .
( left panel ) and @xmath25 ( right panel ) for 0 - 10% au+au collisions at @xmath4 = 200 and 62.4 gev .
the error bars are systematic and statistical errors added in quadrature .
data are compared to model calculations based on coalescence and jet fragmentation in refs .
@xcite.,title="fig : " ] ( left panel ) and @xmath25 ( right panel ) for 0 - 10% au+au collisions at @xmath4 = 200 and 62.4 gev .
the error bars are systematic and statistical errors added in quadrature .
data are compared to model calculations based on coalescence and jet fragmentation in refs .
@xcite.,title="fig : " ] figure [ fig3 ] shows the comparison of @xmath26 , @xmath24 , @xmath25 and @xmath27 ( @xmath28 ) for au+au and cu+cu collisions having similar @xmath29 and @xmath30 at @xmath4 = 200 and 62.4 gev . for the @xmath27 at @xmath4 = 62.4 gev , the @xmath31
is taken to be 36 mb and the @xmath32 data are from the parametrization available from isr data in the ref.@xcite .
we observe , for similar @xmath29 and/or @xmath30 at a given beam energy , the above observables at high @xmath0 have similar values in au+au and cu+cu collisions .
( top left panel ) , @xmath24 , @xmath25 ( top right and bottom left panels ) and @xmath27 ( bottom right panel ) for au+au and cu+cu collisions having similar number of participating nucleons ( @xmath29 ) and number of binary collisions ( @xmath30 ) at @xmath4 = 200 and 62.4 gev .
the error bars are systematic and statistical errors added in quadrature .
the shaded band in the panel showing @xmath33 is the uncertainty due to @xmath30 calculations .
@xmath33 has an additional 25% uncertainty due to the parametrization used for p+p data at @xmath4 = 62.4 gev @xcite . ]
we have presented a study of the energy dependence of the @xmath1 , @xmath2 and @xmath3 production at high @xmath0 from au+au and cu+cu collisions at @xmath4 = 62.4 and 200 gev .
there is a distinct difference in the @xmath0 dependence of @xmath34 for charged pions and protons+anti - protons observed at both energies .
however at higher @xmath0 the values of @xmath17 for baryons and mesons at both energies are similar .
this together with the recent observation of a comparable @xmath25 ratio in 200 gev au+au collisions and @xmath35+au collisions @xcite at high @xmath0 provides data necessary to understand the differences , if any , in the interaction of quarks and gluons and the medium , and through this the energy loss mechanism . the @xmath2/@xmath20 ratio for au+au collisions at @xmath4 = 62.4 gev is higher than corresponding values at @xmath4 = 200 gev in the intermediate @xmath0 range . the case for @xmath3/@xmath21 ratio
is reversed .
the higher value of @xmath2/@xmath20 ratio at intermediate @xmath0 for 62.4 gev compared to 200 gev is qualitatively consistent with quark coalescence models .
however there are quantitative disagreements of data with such models .
we also observe similar values of high @xmath0 particle ratios and @xmath33 in au+au and cu+cu collisions for collisions with similar @xmath29 and/or @xmath30 .
|
we present the incident energy and system size dependence of the @xmath0 spectra for @xmath1 , @xmath2 , and @xmath3 using au+au and cu+cu collisions at @xmath4 = 62.4 and 200 gev in star experiment at rhic . through these measurements in the @xmath0 range of 0.2 @xmath5 @xmath0
@xmath5 10 gev/@xmath6 we conduct a systematic study of the beam energy , system size and particle species dependence of nuclear modification factors and address specific predictions from the quark coalescence models regarding the beam energy dependence of baryon enhancement in the intermediate @xmath0 ( 2 @xmath5 @xmath0 @xmath5 6 gev/@xmath6 ) region .
|
the origin of hot hydrogen - deficient post - agb stars ( spectral types [ wc ] and pg1159 ) is supposed to be a late he - shell flash .
detailed summaries on their spectroscopic characteristics and quantitative analyses , and relevant evolutionary calculations can be found e.g. in werner ( 2001 ) and herwig ( 2001 ) , respectively . in the last white dwarf workshop , held in delaware two years ago , we presented for the first time clear evidence for iron deficiency in a pg1159 star ( miksa 2001 )
. because pg1159 central stars are very hot ( @xmath075000k ) , the metals are highly ionized .
the dominant ionization stage of iron in the line formation region is and most of its lines are located in the fuv spectral region .
contrary to our expectation we were not able to detect any line in the fuse fuv spectrum of the pulsating central star of k1 - 16 , which means that iron is deficient by at least one dex .
this has initiated an effort to utilize archival iue and hst spectra as well as new fuse spectra to look for iron in other pg1159 stars ( miksa 2002 ) .
it turned out that iue and hst spectra do show a hint of iron deficiency in some other objects , but the data quality is not sufficient for reliable quantitative analyses .
the same holds for a number of fuse spectra from pg1159 stars .
however , in two more cases , where fuse spectra with sufficiently high s / n were obtained , we were successful in proving that iron is deficient .
the first one is the central star of ngc7094 , which is classified as a hybrid - pg1159 star , because it exhibits h balmer lines .
no iron lines were detected and we concluded that iron is depleted by at least 1.5 dex ( miksa 2002 ) .
the next case is the central star of abell 78 , which is classified as one of the rare [ wc]pg1159 transition objects . for abell 78
we find an iron deficiency of 1.5 dex from the lack of iron lines as well ( werner 2002a ) . what is the origin of the iron deficiency ?
we think that iron has been transformed to heavier elements .
the high c and o abundances in pg1159 stars result from envelope mixing caused by a late he - shell flash .
this event also modifies the near - solar abundance ratios of iron - peak elements in the envelope by dredging up matter in which s - process elements were built - up by n - capture on @xmath1fe seeds during the agb phase .
this scenario can and will be tested by analyzing the resulting fe / ni abundance ratio , because it is significantly changed in the intershell region in favor of ni by the conversion of @xmath1fe into @xmath2ni .
more quantitative results from nucleosynthesis calculations in appropriate stellar models have been presented recently ( herwig 2002 ) and inclusion of nuclear networks in evolutionary model sequences will become available in the near future .
interestingly , asplund ( 1999 ) have found that in sakurai s object , which is thought to undergo a late he - shell flash , iron is reduced to 0.1 solar and fe / ni@xmath33 .
this and other s - process signatures should also be exhibited by wolf - rayet type central stars and pg1159 stars . and
in fact , latest results confirm that iron deficiency among h - deficient post - agb stars is not restricted to pg1159 stars .
it appears that three wolf - rayet type central stars are iron deficient , too .
grfener ( 2002 ) report a low iron abundance in smp61 , an early type [ wc5 ] central star in the lmc .
its abundance is at least 0.7dex below the lmc metallicity .
crowther ( 2002 ) find evidence for an iron underabundance of 0.30.7dex in the galactic [ wc ] type central stars stars ngc 40 ( [ wc8 ] ) and bd+30@xmath43639 ( [ wc9 ] ) . in this context
it is also interesting to remark that the hot do white dwarf is iron deficient and rich in nickel ( barstow 2000 ) . if this is related to the iron deficiency in pg1159 stars is an open question .
h1504 + 65 is the most extreme pg1159 star , not only because of its extraordinarily high effective temperature , which is close to 200000k .
we have shown that h1504 + 65 is not only hydrogen - deficient but also helium - deficient . from optical spectra
it was concluded that the atmosphere is primarily composed of carbon and oxygen , by equal amounts .
neon lines were detected in soft x - ray spectra taken with the euve satellite and in an optical - uv keck spectrum , and a neon abundance of 25% ( mass fraction ) was derived ( werner & wolff 1999 ) . h1504
+ 65 was observed with chandra on sep .
27 , 2000 , with an integration time of 7 hours .
flux was detected in the range 60160 and the spectral resolution is about 0.1 .
the spectrum displays astonishing details ( fig.1 ) . at first sight
the spectrum appears to be rather noisy , but comparison with model spectra suggests that most absorption features in the observation are probably real .
mg lines are discovered for the first time in h1504 + 65 ( see werner ( 2002b ) for more details ) .
exploratory models including opacities of heavy metals ( ca ni ) were computed .
this introduces a very large number of additional lines , however , identifying individual heavy metal lines will be problematic or even impossible , because the vast majority of lines in the available lists have uncertain wavelength positions .
the mg abundance is of the order 2% , similar to the ne abundance determined earlier .
this allows two alternative interpretations .
either we see a naked c - o white dwarf with 3@xmath5 processed matter , which is , according to evolutionary models , dominated by c and o with an admixture of @xmath6ne and @xmath7 mg of the order 1% .
or we see a naked o - ne - mg white dwarf which has undergone carbon burning .
a comparison with the respective evolutionary models ( ritossa et al . 1999 ) suggests that we could look onto such a naked remnant where @xmath8ne and a mixture of several mg isotopes ( @xmath9mg,@xmath7mg,@xmath10 mg ) should be present .
we have already speculated earlier , that the onset of c burning and subsequent evolutionary processes might explain the uniqueness of h1504 + 65 . in order to confirm this idea
, we will also look for na absorption lines , because the stellar models predict a relatively high amount of @xmath11na in this case .
to conclude , h1504 + 65 might have been one of the `` heavyweight '' intermediate - mass stars ( 9m@xmath12 11m@xmath13 ) which form white dwarfs with electron - degenerate o - ne - mg cores resulting from carbon burning .
this speculation is corroborated by the fact that h1504 + 65 is one of the most massive pg1159 stars known ( [email protected]@xmath15 ) .
asplund , m. , lambert , d.l . , kipper , t. , pollacco , d. , & shetrone , m.d .
1999 , a&a , 343 , 507 barstow , m.a . ,
dreizler , s. , holberg , j.b . , 2000 ,
mnras , 314 , 109 crowther , p.a . ,
abbott , j.b . , hillier , d.j .
, & de marco , o. 2002 , in planetary nebulae , iau symp . 209 , eds . m.dopita et al . ,
in press grfener , g. , hamann , w .- r . ,
& pe~ na , m. 2002 , in planetary nebulae , iau symp . 209 , eds .
m.dopita et al .
, in press herwig , f. 2001 , in low mass wolf - rayet stars : origin and evolution , ed .
r.waters , a.zijlstra , t.blcker , ap&ss , 275 , 15 herwig , f. , lugaro , m. , & werner , k. 2002 , in planetary nebulae , iau symp .
209 , eds .
m.dopita et al . , in press miksa , s. , deetjen , j.l . ,
dreizler , s. , 2001 in white dwarfs , ed .
j.l.provencal , h.l.shipman , j.macdonald , s.goodchild , asp conf .
ser . , 226 , 60 miksa , s. , deetjen , j.l . , dreizler , s. , kruk , j. , rauch , t. , & werner , k. 2002 , a&a , 389 , 953 ritossa , c. , garcia - berro , e. , & iben , i. jr
1999 , apj , 515 , 381 werner , k. 2001 , in low mass wolf - rayet stars : origin and evolution , ed .
r.waters , a.zijlstra , t.blcker , ap&ss , 275 , 27 werner , k. , & wolff , b. 1999 , a&a , 347 , l9 werner , k. , dreizler , s. , koesterke , l. , & kruk , j.w .
2002a , in planetary nebulae , iau symp . 209 , eds .
m.dopita et al .
, in press werner , k. , rauch , t. , barstow , m.a . , & kruk , j.w .
2002b , in exotic stars as challenges to evolution , ed .
c.a.tout , w.van hamme , asp conf .
|
we investigate fuse spectra of three pg1159 stars and do not find any evidence for iron lines . from a comparison with nlte models we conclude a deficiency of 11.5dex .
we speculate that iron was transformed into heavier elements .
a soft x - ray chandra spectrum of the unique h- and he - deficient star h1504 + 65 is analyzed .
we find high neon and magnesium abundances and confirm that h1504 + 65 is the bare core of either a c - o or a o - ne - mg white dwarf .
|
the statistical properties of the ultraviolet variability of active galactic nuclei ( agn ) have been investigated by paltani & courvoisier ( @xcite ) ( hereafter pc94 ) .
this study , which was based on spectra obtained by the _ international ultraviolet explorer _ ( iue ) from the ulda ( _ uniform low dispersion archive _ ) database , has been done in the observer s frame .
in addition to the variability at 2000 , we quantified also the change of variability with the wavelength , which showed that all classes of agn have a larger variability at short wavelength than at long wavelength in the iue domain . in this paper
we shall consider only the objects that have been called in pc94 , i.e. seyfert 1 galaxies , radio - quiet quasars , and low - polarization radio - loud quasars .
the relationship between variability and luminosity in quasars and seyfert galaxies has been investigated by many authors .
most of them found an anti - correlation between these two quantities ( pica & smith @xcite ; cristiani et al .
@xcite ; pc94 ; hook et al .
@xcite ) , while others found a correlation or no correlation at all ( bonoli et al . @xcite ; trvese et al .
@xcite ; giallongo et al .
rediscussion of these last three papers by hook et al .
( @xcite ) seems to show that all the samples are compatible with the existence of an anti - correlation .
all these studies have used estimates of the variability in the observer s frame , as they can be much more easily obtained than those in the objects rest frame .
therefore the correct relationship between variability and luminosity is still unknown .
this relationship is very important , as it can be compared with the predictions of some agn models , in particular the starburst model of terlevich ( @xcite ) . in this paper
we establish this relationship in the objects rest frame , and we discuss some biases that can affect it .
we then check the compatibility between our result and the predictions of these models .
we formalize in the most general way the concept of `` discrete - event models '' , i.e. models where the total light curve is the sum of a series of light curves , each one associated with one `` event '' , whatever its physical nature , and we compare their predictions with the above constraint .
we give the general relationship between variability and luminosity for these models .
we study two different ways by which this relationship can be modified . in the first one we assume that some of the event parameters may actually depend on the luminosity . in the second one
we examine the effects of the introduction of interdependence between the events .
in this paper , as in pc94 , we use the iue spectra from the ulda version 4.0 database .
this database contains nearly every low - dispersion spectrum obtained by the iue satellite before 1992 .
the period covered by the ulda archive is therefore about 14 years .
the wavelength range covered by these spectra , combining short- and long - wavelength cameras , is from 1200 to 3200 .
the large number of repeated observations and the spectral coverage of the iue satellite make this satellite ideal for the determination of the agn variability properties . in pc94 we estimated for each object the wavelength - dependent variability , @xmath2 .
in all this paper , as in pc94 , `` variability '' should be understood as the standard deviation of the flux divided by the mean flux . for each object we approximated @xmath2 by a linear relationship : @xmath3 @xmath4 is expressed in percents of the mean flux and @xmath5 , the slope of @xmath2 , is expressed in percents of the mean flux per 1000 ( cf .
fig . 1 of pc94 ) .
the values of @xmath4 and @xmath5 can be found in the appendix of pc94 .
the definition of these parameters has only a descriptive purpose and does not have any predictive value outside the wavelength range studied . using these two parameters , we can derive the agn variability at a given wavelength , 1250 in our case , in the rest frame of the objects : @xmath6 we feel that it is important not to extrapolate the variability outside the iue wavelength range .
we therefore consider in this paper only the objects that have a redshift smaller than 1.3 .
it is important to estimate the uncertainty on @xmath7 , as we shall be concerned in this paper with the relationship between variability and another parameter , namely the luminosity .
the uncertainty on @xmath7 has two origins .
first it is due to the propagation of the uncertainties on @xmath8 , @xmath4 , and @xmath5 from pc94 .
as explained in pc94 , the two last uncertainties are the uncertainties on the least - squares linear regression and do not take into account the effect of the sampling .
a second term must therefore be included to allow for the effect of the sampling .
we propose here a method that is an improvement of the discussion of sect . 3.2 in pc94 .
the first step is the simulation of a light curve that has variability properties comparable to those of our objects .
we assume that the agn light curves are characterized by a broad fourier power spectrum which follows a power - law with an index between @xmath9 ( flicker noise ) and @xmath10 ( random walk ) .
these has been verified for at least two : 3c 273 ( kunkel @xcite ) and ngc 4151 ( longo et al .
@xcite ) ( see also press @xcite ) .
an example of a light curve simulated by a `` random walk '' can be seen in fig . 1 of pc94 . we rescale these simulated light curves to obtain a variability comparable to the ones observed in our objects .
then we draw @xmath11 epochs from an uniform distribution on the period covered by ulda and compare the variability estimated with these @xmath11 points with the variability of the simulated light curve . repeating this last step many times
, we can estimate the uncertainty produced by the sampling from the dispersion of the estimates of the variability .
as the indices are smaller or equal to -1 , a low - frequency cut - off has to be chosen if one wants to avoid infinite variability . on the basis of archival data of 3c 273 ( angione & smith @xcite )
, it appears that variability on time scales longer than 20 years is not dominant .
if this is valid for all agn , a 14-year period should contain almost all the power of the fourier spectrum .
the effect of variations on longer time scales and of time dilation in high - redshift objects are discussed in sect .
[ verylong ] and [ tim - dil ] respectively . taking 3 values for the spectral index of the power spectrum ( @xmath9 , @xmath12 , and @xmath10 ) , 3 values for the low - frequency cut - off ( ( 5 yr)@xmath13 , ( 10 yr)@xmath13 , and ( 20 yr)@xmath13 ) , and 3 values for the relative variability ( 0.05 , 0.1 and 0.5 ) , we parametrize the relative uncertainty on @xmath14 , @xmath15 , as a function of @xmath11 : @xmath16 ( where @xmath17 is the uncertainty for @xmath18 ) , which produces good approximations .
we normalize @xmath15 at @xmath18 because this value is central and is little affected by small deviations of @xmath19 .
it appears from the simulations that none of the 3 parameters modifies significantly the value of @xmath19 , and @xmath20 reproduces quite well the decrease of @xmath15 , at least for @xmath21 . on the other hand
, @xmath17 depends on the 3 parameters , but the differences are only a few percents .
@xmath17 increases when the index increases , the low - frequency cut - off decreases , or the variability increases .
we therefore use the following empirical relation for the uncertainty on the relative variability due to the sampling in the worst case , i.e. low - frequency cut - off@xmath22(20 yr)@xmath13 , and index@xmath23 ( flicker noise ) : @xmath24 the best case makes @xmath17 become @xmath25 . in practice , as the number of spectra is usually different in the short- and long - wavelength ranges , @xmath11 will be the larger of the two numbers .
for the few cases where the number of spectra is larger than 50 , we use @xmath26 .
it is important to note that the estimates of the variability follow rather well a gaussian distribution as soon as @xmath11 is larger or equal to 4 ( the distribution is already marginally acceptably gaussian for @xmath27 ) .
thus we can estimate the total uncertainty on @xmath7 by simply adding the variances of the two sources of dispersion .
figure [ blum ] shows the variability @xmath7 as a function of the luminosity for all the objects classified as in pc94 ( the objects and their properties are given in the appendix of pc94 ) .
the spearman s correlation coefficient is equal to -0.45 .
the probability to get such a high correlation coefficient for 68 couples drawn from two uncorrelated parent populations is about @xmath28 .
we note that spearman s test is valid even if the variables have uncertainties or intrinsic scatter .
a real correlation may be destroyed by the dispersion , but the probability that the dispersion generates a spurious correlation is very low .
therefore , with better measurements , the correlation coefficient should in principle increase ( in absolute value ) .
it appears therefore unquestionable that a trend exists , and this trend is in the sense that the variability decreases when the luminosity increases . to quantify this trend , one must choose an analytical formulation . because the very large dynamic of the @xmath29 axis ( @xmath30 ) , the most natural one is a power - law relation in the form :
@xmath31 where @xmath32 is the average luminosity density of the object at 1250 in the rest - frame of the object .
the index @xmath33 of the trend can in principle be estimated . however , because of the important uncertainties , one must be careful about the selection of the method .
as we determined both @xmath29- and @xmath34-axis uncertainties , we used a @xmath35 method that takes into account the uncertainties on both coordinates ( press et al .
this method is valid only for linear ( affine ) relationships ; therefore we fit the logarithm of the values .
unfortunately , we have to estimate the variances of the logarithms of the distributions , which means that the uncertainties are not gaussian anymore .
we nevertheless assume that they follow closely enough gaussian distributions in order to apply the @xmath35 test .
we obtain @xmath36 .
the probability that a linear relationship really holds is of the order of @xmath37 ( @xmath38 with 66 degrees of freedom ) . as our uncertainties are probably overestimated , some intrinsic scatter around this relationship does exist , but a large part of the dispersion can be accounted for by the uncertainties . the linear function with a slope fixed to @xmath1
that minimize the @xmath35 has been plotted on fig .
it is clearly unadapted to the data ( its @xmath35 is larger than 1000 ) .
the value of the slope obtained using the usual least - squares method ( which neglects the uncertainties on both axes ) is @xmath39 , which is very close .
the scatter makes the uncertainty on the slope of the relationship meaningless .
fortunately a new method has appeared that is not only applicable when uncertainties on both axes exist , but also when there is some intrinsic scatter .
this method , called bces ( akritas & bershady @xcite ) , gives a slope @xmath40 .
the agreement between all these methods reinforces the confidence in the value , and we use in the following : @xmath41 the bces method is , as the authors claim , very conservative . although @xmath42 is formally possible , spearman s test , which is much more efficient in detecting the existence of a slope , clearly rules out this hypothesis .
it is interesting to note that hook et al .
( @xcite ) have found indices of @xmath43 ( their model a ) and @xmath44 ( their model b ) for the same relationship , which are both compatible with our result ( but see sect .
[ vrrel ] ) .
their quasar sample extends far beyond our sample in redshift and in luminosity and does not comprise any seyfert galaxy . moreover , as we discuss in the next section , they do not allow for several biases .
we have seen in sect .
[ covarlum ] that the slope of the variability
luminosity relationship , although statistically different from 0 , is small . therefore , we can ask whether this trend could be due to a bias . on the other hand ,
it is also important to ask whether the slope of the relationship might be lessened by systematic effects . the ulda 4.0 data span about 14 years .
it means that variability on time scales longer than @xmath45 years will not contribute ( or only partly ) to our estimation of @xmath2 .
if only low - luminosity objects vary on very long time scales , the observed @xmath46 relationship would be flatter than the intrinsic one . on the other hand , the decrease of variability with luminosity
may be introduced by bias if only high - luminosity objects vary on very long time scales .
very - long - term variability can have two manifestations : either slow , steady variations or short , intense flares occuring with a rate of around 1 per century , or less . in any case , this very - long - term variability implies that the fourier power spectrum of the agn light curves should contain a non - negligible amount of power at frequencies smaller than @xmath47 . we can estimate this power for two limiting cases _ assuming _ a given intrinsic @xmath46 relationship : * the variability is independent of the luminosity . in this case
the more luminous the object , the larger the unobserved proportion of power in the fourier spectrum at very small frequencies . assuming that only variability on short time scales exists in low - luminosity objects , it implies that short - time - scale variability represents only 40 % in high - luminosity objects .
the variability of 3c 273 from optical archival data since 1887 ( angione & smith @xcite ) is about 16 % .
its variability on a 10-year basis is 12 % ( work in progress ) . in the fourier spectra presented by angione & smith ( @xcite ) , much power seems to be contained between ( 20 yr)@xmath13 and ( 10 yr)@xmath13 , which could easily explain the difference , but the increase in variability is much smaller than the required factor 2.5 .
the power contained between ( 100 yr)@xmath13 and ( 20 yr)@xmath13 seems negligible . even though it is _ a priori _ possible that still much power exists at frequencies below ( 100 yr)@xmath13 , the small increase of power between ( 100 yr)@xmath13 and ( 20 yr)@xmath13 , together with the necessity of the disappearance of this power in low - luminosity objects , makes this possibility rather improbable . *
the intrinsic @xmath46 relationship has a @xmath1 slope , which is the slope predicted by many discrete - event models ( see sect . [ discrete ] ) . in this case
the variability of low - luminosity objects must have been underestimated by a factor about 30 .
assuming that the variability of high - luminosity objects is correct , which means that only variability on short time scales exists in these objects , the luminosity of low - luminosity objects should be dominated in average by very - long - term variations . as short - term variability
is not much smaller than 100 % in most low - luminosity objects , it would imply that all these objects are in phases where the very - long - term variations have a very small luminosity .
this is statistically impossible if the very - long - term variations are slow and steady .
short intense flares can also be dismissed , as a sufficient variability can be attained only if the average luminosity is dominated by the very rare flares by a large factor .
such flares have never been observed .
taking into account the intense monitoring of several seyfert 1 galaxies , the hypothetical flaring state can be active only for a very small fraction of the time .
consequently , the flare intensity should be extremely large , should have been detected in the same way as supernovae are detected . in conclusion , although we can not completely exclude any important effect on the variability
luminosity relationship , it appears more probable that such biases are marginal .
but if the variability luminosity relationship is affected , it is certainly in the sense that the trend is artificially produced rather than in the sense that the @xmath1 relationship is flattened .
if light curves of agn are stochastic time series that have fourier power spectra well described by a power - law with indices between @xmath9 and @xmath10 , the total variability of the objects increases with the length of the sampling until this length becomes comparable to the inverse of the frequency of the low - frequency cut - off . because of the time dilation , high - redshift objects are potentially observed for a smaller time than low - redshift ones .
let us assume that the `` standard '' index is @xmath10 , which corresponds to a `` random walk '' . if @xmath48 is such a time series , its structure function @xmath49 ( rutman @xcite ; simonetti et al .
@xcite ) is linear : @xmath50 .
larger values of the power - spectrum index make @xmath49 increase more slowly .
the value of the structure function is also related to the variance of the light curve : @xmath51 where @xmath52 is the variance of a portion of the time series @xmath29 , whose length is @xmath53 .
however , because of the redshift of the sources , one does not measure the correct length of the time series @xmath53 , but the observer s length , which is @xmath54 , where @xmath8 is the redshift .
therefore , assuming that the observer s lengthes of the time series are all identical , one has to modify the observed variability : @xmath55 which gives : @xmath56 this correction is an upper - limit , because it assumes that the power spectrum extends down to frequencies much smaller than ( 14 yr)@xmath13 .
if one applies this correction to our data , the anticorrelation still holds with a confidence of 99.8 % , and the slope obtained by the bces method is @xmath57 .
therefore , it seems that the decrease of the variability with the luminosity can not be entirely explained by this effect .
luminosity and redshift are always very well correlated in flux - limited samples , since only the most luminous objects can be observed at large redshift .
this leads to a correlation between variability and redshift .
the question that arises is to determine whether this correlation is entirely due to the bias , or whether there also exists an intrinsic correlation .
hook et al .
( @xcite ) , using subsamples with small luminosity range , have concluded that the first alternative is the correct one .
two effects can produce a correlation between variability and redshift .
the first is evolution ; unfortunately , we are much too far from a clear picture of how agn work to guess anything about its effect , if any , on the variability redshift correlation .
the second effect is much clearer ; in pc94 , we showed that the variability increases with the frequency .
as the observations are usually performed at a wavelength fixed in the observer s frame , this should naturally lead to a positive correlation between variability and redshift .
di clemente et al .
( @xcite ) found that this effect alone can account for the positive correlation between variability and redshift , which means that there is no need to invoke evolution effects . to allow for this effect
, cid fernandes et al .
( @xcite ) have fitted the data of hook et al .
( @xcite ) with a function of two variables , the luminosity and the redshift , and three parameters , @xmath58 , @xmath59 , and @xmath60 : @xmath61 where @xmath62 is the mean luminosity in the optical b band and @xmath8 is the redshift .
the parameter @xmath58 is is equivalent to our parameter @xmath33 ; @xmath59 takes into account the increase of variability with the frequency ; and @xmath60 is a normalization parameter .
although evolution effects could in principle affect parameter @xmath59 , this is not considered by cid fernandes et al .
( @xcite ) , because their preferred model makes use of independent events , identical for all objects ( see sect .
[ discrete ] ) .
the results of their study are : @xmath63 and @xmath64 ( in one particular case ; however other cases are qualitatively identical ) .
it is in clear contradiction with the results cited above , in particular ours .
the explanation of the difference between this result and the one of , for instance , hook et al .
( @xcite ) lies essentially in two points : 1 .
cid fernandes et al .
( @xcite ) take into account the fact that agn vary more at shorter wavelengths .
hook et al .
( @xcite ) neglect entirely this effect after checking that it is not important in their sample .
2 . cid fernandes et al .
( @xcite ) estimate the uncertainties on the luminosity , and take them into account in the fits .
la franca et al .
( @xcite ) have shown that the slopes are biased towards a smaller value when the errors on the @xmath29 axis are discarded . since
our result is very close from the one of hook et al .
( @xcite ) , who neglect these two biases , further discussion is required .
first , we note that the increase of the @xmath59 parameter has an effect very close to the decrease of @xmath58 .
this implies that solutions with @xmath65 are indeed allowed by cid fernandes et al .
( @xcite ) , but it requires large values of @xmath59 ( @xmath66 ) .
although such values may be allowed in some cases , it can not be guaranteed that most objects follow this behaviour .
the determination of the rest - frame variability removes the difficulty encountered by cid fernandes et al .
( @xcite ) about the value of @xmath59 . in our case ,
no hypothesis , no extrapolation is needed .
as evolution is probably not a problem ( the more so as all our objects have redshift smaller than 1.3 ) , no correlation between variability and redshift is expected , apart , perhaps , from the time - dilation effect discussed in sect .
[ tim - dil ]
. we can object to the second point that we also took into account the uncertainties on the @xmath29 axis to obtain the slope . on the other hand
cid fernandes et al .
( @xcite ) did not estimate the uncertainties on the variability , which , in our case , are much larger than those on the luminosity .
they obtain steep values for the variability
luminosity relationship only in the cases where the uncertainty on the luminosity is very large ( which is , by the way , an excellent illustration of the bias described in la franca et al .
( @xcite ) ) .
moreover , only the bces method takes correctly into account the existence of intrinsic scatter in the data .
once again , we took advantage of the spectral capabilities of iue : while cid fernandes et al .
( @xcite ) have to derive @xmath67-correction based on unknown spectral indices , our luminosities are automatically calculated at 1250 in the rest frame of the objects , again without any hypothesis or extrapolation .
we note that the @xmath1 slope is compatible with our data at the 1 @xmath14 level if the uncertainties on the luminosity are multiplied by a factor 18 ( @xmath68 ) , which is clearly an overestimate ( moreover , the bces method has very conservative uncertainties ) . as a consequence
we infer that the difference in the result is due mostly to the assumptions that they had to make to correct from the wavelength dependence of the variability and to obtain rest - frame luminosities .
it is very convenient to think of the variability as due to the superimposition of perturbations on a constant source . the luminosity density at frequency @xmath69
can then be written : @xmath70 where @xmath71 is the constant state of the object and @xmath72 is the light curve produced by one event ( the index @xmath73 indicates that the event light curve may be different for each event ) , and @xmath74 is the `` birth time '' of the @xmath75 event .
it is a generalization of a shot - noise process .
this representation can apply to the blue - bump of the , as the variations are quasi - simultaneous from the ultraviolet to the optical ( courvoisier & clavel @xcite ) .
moreover it is very general , and can be a reasonable approximation of many models , from accretion disk models , e.g. the `` blob '' model ( haardt et al .
@xcite ) or shocks in the disk ( chakrabarti & wiita @xcite ) , to supernova explosions ( terlevich @xcite ) and relativistic star collisions ( courvoisier et al . @xcite ) . the interesting point of the discrete - event models is that the variability properties of a set of object is subject to strong constraints as soon as one puts limitations on eq .
( [ eq - dis_ev ] ) . to demonstrate analytically these constraints
, we develop the following formalism : the mean luminosity density @xmath76 , which can be written ( in the following we omit the `` @xmath69 '' index , the terms `` luminosity '' and `` energy '' must be understood as `` luminosity density '' and `` energy density '' , and we assume that the studied systems are completely stationnary ) : @xmath77 where @xmath78 is the event mean rate , @xmath79 is the total energy released by the @xmath75 event ( at frequency @xmath69 ) and @xmath80 is the mean of @xmath81 over all @xmath73 s .
( [ eq - mean ] ) is valid even if the @xmath82 differ from each other .
it is much more complicated to obtain the variance of @xmath83 .
its exact definition is : @xmath84 using the hypothesis that the light curve is given by eq .
( [ eq - dis_ev ] ) , it is possible to transform this equation .
the analytical development is made in appendix a and the result is given in eq .
( [ eq - an_var_all ] ) .
if one makes the assumption that the event rate is independent of the parameters defining the events , one finds that ( from eq .
( [ eq - an_var_ind ] ) ) : @xmath85 the relative variability , i.e. the ratio between the standard deviation ( the square root of the variance ) and the mean , is therefore : @xmath86 the strongest limiting case on eq .
( [ eq - dis_ev ] ) has already been discussed by pica & smith ( @xcite ) : the whole luminosity is due to the events ( @xmath87 ) and all events are independent and identical .
( [ eq - sigma ] ) shows that @xmath88 is proportional to @xmath89 , i.e. proportional to @xmath90 ( by eq .
( [ eq - mean ] ) ) .
however , eq . ( [ eq - sigma ] ) is much more general ; in particular the events may be completely arbitrary , and the @xmath1 index will still be preserved .
@xmath87 is not even necessary .
the @xmath1 index is also obtained if @xmath91 is a fixed fraction of the total luminosity ( @xmath92 ) . moreover ,
if @xmath91 is constant for all objects ( e.g. if @xmath91 is the contribution of the underlying galaxy ) , we still find the @xmath1 index as soon as @xmath91 is negligible compared to the events luminosity . in conclusion
, we see that the only requirements to preserve the index @xmath1 is that the parameters of the events are drawn from the same distributions in all objects ( i.e. the distributions of the parameters are independent of @xmath78 ) .
pica & smith ( @xcite ) concluded therefore that such models can be rejected independently of the nature of the events ( although they did not prove this rigourously ) .
they have found that the @xmath1 index could be dismissed with a confidence larger than 99 % . pc94 and hook et al .
( @xcite ) have reached a similar conclusion . however in all these studies
the variability have been made in the observer s frame . in this paper ,
the variability used to determine the relationship with the luminosity is corrected for the redshift effect , and here again the value @xmath1 can be rejected with a large confidence .
terlevich ( @xcite ) proposed that the agn could be powered by independent supernova explosions . in this model , the events ( the supernovae ) are identical and independent , and @xmath93 is satisfied , because a fixed fraction of the total luminosity is emitted by normal ( constant ) stars ( @xmath29 is larger than 0 , but depends strongly , of course , on the considered wavelength ) .
as we have seen , the index of the relationship between luminosity and variability remains equal to @xmath1 .
even if a general trend is predicted , we see that much room is left for the observed intrinsic scatter in the terms inside the integrals in eq .
( [ eq - an_var_ind ] ) .
for instance the distributions of the luminosities of the events may differ from object to object without being necessarily related to the total luminosity of the object .
as the observations clearly prohibit the so universal @xmath1 index , we could ask whether the release of one of the two properties of the events ( `` identity of the distributions of the event parameters '' and `` independence '' ) might produce the observed behaviour . the variability
luminosity relationship can be modified to match any given relationship if one adjusts in an _ ad - hoc _ way the distributions of the event parameters .
we examine only two very simple cases that can be intuitively expected in many discrete - event models .
a very simple modification that breaks the `` identity of the distributions of the event parameters '' assumption is that the energy released by the events scales with the total luminosity of the object , while the events are still independent from each other .
we can estimate the effect of such a scaling on the variability
luminosity relationship using the notation of eq .
( [ eq - scal_not ] ) and the results of eqs ( [ eq - mean_scal ] ) and ( [ eq - var_scal ] ) ( see appendix b ) in the case @xmath94 , @xmath95 .
the scaling is written @xmath96 , which gives @xmath97 .
therefore eq .
( [ eq - sigma ] ) becomes : @xmath98 if no correlation exists between variability and luminosity ( @xmath42 ) , then we must have : @xmath99 .
if the @xmath100 index is correct , one obtains : @xmath101 this scaling can not be applied to models based on supernovae , as the energy released by a supernova is essentially fixed . on the other hand ,
it is reasonable to assume that the luminosity of an event of unspecified nature related to the existence of a black hole scales in a way or in another with its mass .
for instance , courvoisier et al .
( @xcite ) have proposed a model where such a scaling with luminosity could be rather natural . in this model ,
the events are collisions between stars in a dense stellar cluster surrounding a supermassive black hole , which liberates the stars relativistic kinetic energy . for a given cluster ,
an increase of the black - hole mass will increase the energy liberated by the events .
generally speaking , however , the value of @xmath102 shows that the event luminosity is not trivially related to the total luminosity .
this may be a major difficulty for the discrete - event models .
we call `` life time '' of an event any time that is mathematically associated with the event .
for instance , the life time of a `` box '' event is the duration of the non - zero state ; for an semi - exponential event , the life time may be the @xmath103-folding time .
we may think that the event life time increases with the total luminosity .
this may appear naturally in many models .
for instance , in case of reprocessing on an accretion disk , the size of the reprocessing area ( contributing at a given frequency ) very certainly increases with the luminosity of the x - ray source(s ) .
we call the possibility to modify the event life time the `` stretching '' of the events .
we introduce in the mathematical formulation of the events a stretching factor @xmath104 ( corresponding to `` @xmath105 '' in eq .
( [ eq - scal_not ] ) ) .
we assume that @xmath104 scales with the total luminosity of the object , i.e. @xmath106 ( see appendix b ) . from eqs ( [ eq - mean_scal ] ) and ( [ eq - var_scal ] )
we obtain ( valid if @xmath107 ) : @xmath108 if @xmath42 , we must have @xmath109 ; with our value for @xmath33 we find : @xmath110 these values are not satisfactory , because one would rather expect that the stretching factor increases with the luminosity , and not decreases .
it appears therefore that this mechanism ( alone , at least ) can not explain the variability luminosity relationship .
we can also combine the scaling of the event luminosity and the one of the event stretching .
again , the justification of these scalings is that they seem natural to us .
eqs ( [ eq - mean_scal ] ) and ( [ eq - var_scal ] ) give here : @xmath111 if we assume that @xmath112 , which means that the stretching factor of the events is proportional to the total luminosity , our result requires : @xmath113 or , if @xmath99 , i.e. the event luminosity is proportional to the total luminosity : @xmath114 if we impose @xmath115 and @xmath116 , which is physically the most appropriate , we find that @xmath102 must be @xmath117 , because @xmath118 . if @xmath42 , we must have @xmath119 , and @xmath120 .
the models are at the moment not sufficiently developed to predict any value for @xmath102 and @xmath121 .
we can however remark that the constraints on @xmath102 and @xmath121 are in no case intuitively `` natural '' .
it has often been argued that a different relationship between variability and luminosity could be obtained if the events were not independent from each other .
for instance , one can imagine a cluster of explosive charges .
the charges may explode spontaneously ( producing an `` event '' ) , or the explosion may be triggered by the close explosion of another charge .
it is easy to imagine a mathematically analogous situation in agn . in this picture , each spontaneous event ( not triggered by another event ) is followed by a sequence of `` child '' events . therefore one can replace the spontaneous event and its `` child '' events by one `` super - event '' .
the normalization and shape of this `` super - event '' are not constant , but are drawn from complex distributions .
it appears therefore from the discussion in sect .
[ proc_poiss ] that the @xmath1 index can not be avoided by any process following this picture .
for instance , if all the events are independent and if the `` parent '' event has a probability @xmath122 to produce a simultaneous `` child '' event , which in turn has a probability @xmath122 to produce another `` child '' event , and so on , the variability at a given total luminosity ( if @xmath87 ) is multiplied by : @xmath123 .
the reason why the @xmath1 index always holds in the processes dicussed above is that the process remains poissonian in essence .
however no clear physical picture of non - poisson process can be proposed at the moment .
we can nevertheless propose a mathematical non - poisson process : in a system where events occur according to the poisson law , the probability @xmath124 that no event occurs between @xmath125 and @xmath126 follows the equation : @xmath127 with the boundary condition @xmath128 , @xmath129 being the mean number of events between @xmath125 and @xmath130 .
the solution of eq .
( [ eq - poiss ] ) is simply : @xmath131 we can modify eq .
( [ eq - poiss ] ) , so that the mean time without event decreases when new events occur ; this reflects the interdependence of the events .
we write : @xmath132 where @xmath133 is the `` response '' to the occurence of an event whatever its origin , @xmath11 is the number of events that have already happened , and @xmath134 is the time distribution of the influence of an event , with @xmath135 .
@xmath133 should be smaller than 1 , as a larger value would produce a diverging `` chain reaction '' .
it is possible to solve eq .
( [ eq - non_poiss ] ) numerically for each new event and to draw lists of successive events for a given choice of @xmath129 and @xmath133 .
if we choose an event light curve , we can construct the total light curve , and deduce the luminosity and the variability of the virtual object .
in all the simulations , the distribution function @xmath134 has been taken to be uniform between 0 and 1 ( arbitrary ) unit of time ( @xmath136 ) and the event light curve was a simple box function . for a set of @xmath133 chosen between 0 and 0.96 , we calculated different light curves , making @xmath129 , the rate of independent events , vary from @xmath37 to as many events per @xmath136 as possible ( the limitation is due to the calculation time , which increases with the rate of events ) . examples of light curves are shown in fig .
[ depclum ] .
figure [ simvarlum ] shows the variability luminosity relationship obtained from the simulations for different values of @xmath133 . in all curves
the behaviour is comparable ; one can distinguish 3 regimes : * the _ sub - critical regime _ occurs when the mean rate of events is small . the variability
luminosity relationship has the slope predicted by pure poisson statistics , independently of @xmath133 .
the events are too rare to affect significantly eq .
( [ eq - non_poiss ] ) .
the light curves ( fig .
[ depclum]a ) do not show clear structures ( as is expected from pure poisson process ) .
* the _ critical regime _ occurs when the mean rate of events is around 1 per @xmath136 ( slightly dependent on @xmath133 ) .
the variability decreases less rapidly , or even increases for large values of @xmath133 .
the larger @xmath133 , the larger the deviation from the @xmath1 index . in this regime ,
the event density is right in the range that produces `` chain reaction '' .
the light curves show very characteristic , flare - like structures due to the interdependence of the events ( e.g. around @xmath137 @xmath136 in fig .
[ depclum]b ) . *
the _ super - critical regime _ occurs when the mean rate of events increases further .
the variability
luminosity relationship has again the slope predicted by pure poisson statistics , but with a larger normalization than in the sub - critical regime .
it seems that the `` chain reaction '' is saturated by the large event density , and the light curves ( fig .
[ depclum]c ) have essentially lost the characteristic structures , probably because of the `` overcrowding '' of the events .
the increase of variability in this regime as a function of @xmath138 is compatible with the multiplication by a factor @xmath139 . even though this model can produce a variability
luminosity relationship different from the poisson case , none of the curves in fig .
[ simvarlum ] can be adjusted to the data in fig .
moreover , the physical feasability of this model can not be asserted .
other mathematical formulations might yield more appropriate results .
however , the physical feasability might be a major difficulty in most ( if not all ) of them , because ( pseudo-)poisson processes appear much more natural .
the trend between the variability and the luminosity seen by many authors is confirmed when one estimates the rest - frame variability .
the most probable index of the relationship is @xmath0 .
it is compatible with results obtained using high - luminosity quasars .
this is an additional argument for the continuity between the seyfert galaxies and the most powerful quasars .
we have investigated the effects of several biases , and we consider our result as very probably correct .
an intrinsic @xmath140 index , which is what is generally expected from discrete - event models , would be possible only if there exists very large and yet undiscovered biases ; but this seems highly unlikely .
the other extreme , which is that the variability does not depend on the luminosity , is more plausible .
however the difficulty in producing such a relationship with discrete events is not significantly alleviated .
the easiest way to produce a relationship with an index different from @xmath1 is to assume that the average event luminosity increases with luminosity . in this case
, it would mean that the event luminosity must increase only about 15 % more slowly than the total luminosity .
the scaling of the event stretching factor may also contribute to the relationship .
the relationships given here may provide useful constraints for any kind of discrete - event model .
the possibility that the birth time of the events does not follow a poisson distribution has also been investigated and can not be completely excluded .
however the system that we have explored can not ( alone ) produce the correct relationship .
anyway a non - poisson system appears somehow unnatural , and the physical situation at its origin might be difficult to find . the predictions of physical models that can be expressed in terms of discrete events can be compared quantitatively with the variability luminosity relationship using eq .
( [ eq - an_var_ind ] ) .
but , it appears to us that discrete - event models do not satisfy the variability
luminosity relationship in a natural way .
this difficulty possibly means that this kind of model does not apply to the active galactic nuclei , and , therefore , that eq .
( [ eq - dis_ev ] ) is not valid .
other source of variability ( not considering the possibility that variations are extrinsic , which seems very unlikely ) could be global effects , which do not have a simple general analytical formulation , in opposition to the local processes discussed here . the careful and critical reading of the referee has contributed to improve significantly the mathematical aspects of this paper .
sp gratefully thanks the system managers of the geneva observatory and of the isdc for having allowed him to run the simulations .
sp also aknowledges a grant from the swiss national science foundation .
akritas m.g . &
bershady m.a .
, 1996 , apj 470 , 706 angione r.j . &
smith h.j .
, 1985 , aj 90 , 2474 bonoli f. , braccesi a. , federici l. , zitelli v. & formiggini l. , 1979 , a&a 35 , 391 chakrabarti s.k . & wiita p.j . , 1993 , apj 411 , 602 cid fernandes r. , aretxaga i. & terlevich r. , 1996 , mnras in press courvoisier t.j .- l .
& clavel j. , 1991 , a&a 248,389 courvoisier t.j .- l . , paltani s. & walter r. , 1996 , a&a 308 , l17 cristiani s. , vio r. & andreani p. , 1990
, aj 100 , 56 di clemente a. , giallongo e. , natali g. , trvese d. & vagnetti f. , 1996 , submitted giallongo e. , trvese d. & vagnetti f. , 1991 , apj 377 , 345 haardt f. , maraschi l. & ghisellini g. , 1994 , apj 432 , l95 hook i.m . , mcmahon r.g . , boyle b.j . &
irwin m.j . , 1994 ,
mnras 268 , 305 kunkel w.e .
, 1972 , aj 72 , 1341 la franca f. , franceschini a. , cristiani s. & vio r. , 1995 , a&a 299 , 19 longo g. , vio r. , paura p. , provenzale a. & rifatto a. , 1996 , a&a 312 , 424 paltani s. & courvoisier t.j .- l . , 1994 , a&a 291 , 74
pc94 papoulis a. , 1991 , probability , random variables , and stochastic processes , @xmath141 edition , mcgraw - hill pica a.j . & smith a.g .
, 1983 , apj 272 , 11 press w.h . , 1978 , comments astrophys . , 7 , 103 press w.h . , vetterling w.t . , teukolsky s.a .
& flannery b.p . , 1992 , numerical recipes , @xmath142 edition .
cambridge university press rutman j. , 1978 , proc .
ieee 66 , 1048 simonetti j.h . , cordes j.m . &
heeschen d.s . , 1985 , apj 296 , 46 terlevich r. , 1992 .
in : filipenko a.v .
( ed . ) , relationships between activ galactic nuclei and starburst galaxies .
asp conference series vol .
31 , san francisco trvese d. , pitella g. , kron r.g . & bershady m. , 1989 , aj 98 , 108
the variance of the generalized shot - noise process given in eq .
( [ eq - dis_ev ] ) can be calculated using campbell s theorem ( e.g. papoulis @xcite ) , which states that if : @xmath143 where the points @xmath74 follow a poisson distribution with a mean rate @xmath144 , then the variance of @xmath145 is given by : @xmath146 in our case , everything happens as if we have a superimposition of an infinite ( possibly continuous ) number of shot noises given by the relationship : @xmath147 ( we neglect the constant term , because it does not appear in the variance ) .
the @xmath148 vector is the list of parameters that define the event ; @xmath148 is an element of the @xmath149 space , of arbitrary dimension , and follows a distribution @xmath150 with @xmath151 .
the average rate of these shot noises are @xmath152 . for each @xmath148 ,
the variance of the shot noise is : @xmath153 therefore the variance of the total light curve is the sum ( integration ) of the variance of all shot noises : @xmath154 @xmath155 if one make the assumption that the rates @xmath156 are independent of @xmath148 , eq . ( [ eq - an_var_all ] ) becomes : @xmath157 which gives eq .
( [ eq - var ] ) .
eq . ( [ eq - an_var_all ] ) can be developed further if we assign physical meanings to some components of @xmath148 .
the normalization ( i.e. the total energy liberated by an event ) and the stretching factor are respectively the first and the second components of @xmath148
. we can write : @xmath158 where @xmath159 .
we impose @xmath160 , so that we have @xmath161 .
the reason why @xmath162 is multiplied by @xmath163 is that it makes @xmath164 be independent of @xmath105 ( because @xmath165 ) .
let us make the additional assumption that @xmath166 and @xmath105 are independent of each other and of the other @xmath167 s .
let us write @xmath168 , @xmath169 , and @xmath170 the spaces from which are drawn @xmath166 , @xmath105 , and @xmath171 respectively .
this makes eq .
( [ eq - an_var_ind ] ) become : @xmath172 @xmath173 @xmath174 @xmath175 @xmath176 @xmath177 @xmath178 @xmath179 this relationship is used to derived the scaling properties in sect .
[ nonid ] . to obtain the results shown in this section ,
we introduce two random variables , @xmath180 and @xmath181 , with the relations : @xmath182 and @xmath183 ( we call @xmath184 the multiple integral ) .
the above equation becomes : @xmath185 with the same substitutions , the mean luminosity becomes : @xmath186 eqs .
( [ eq - mean_scal ] ) and ( [ eq - var_scal ] ) give directly eqs .
( [ eq - scal_lum ] ) , ( [ eq - scal_dur ] ) , and ( [ eq - scal_both ] ) .
|
a study of the ultraviolet continuum variability ( paltani & courvoisier @xcite ) has shown that the relative variability of quasars and seyfert galaxies decreases when the luminosity increases .
the spectral information included in the iue spectra allows us to study this dependence in the rest frame of the objects .
the trend is strengthened by the general property that active galactic nuclei vary more at short wavelengths than at long wavelengths in the ultraviolet domain .
the scatter observed in all other studies is still present .
an important part of this scatter may however be explained if one tries to estimate the uncertainties on the variability due to the sampling .
we discuss the variability using the concept of discrete events .
the trend between variability and luminosity is described by a power - law with an index @xmath0 , which is incompatible with the power - law of index @xmath1 predicted by the most general discrete - event models in which the change in average luminosity is due to differences in average event rates exclusively .
several biases are investigated , but we conclude that the @xmath1 index is definitely inconsistent with the data .
a flat relationship is however possible , if some bias has been underestimated .
we propose different ways whereby discrete events may produce a different variability
luminosity relationship : by changing the luminosity or the life time of the events , or by introducing interdependence between the events .
the latter possibility can not produce a satisfactory relationship .
using the former possibilities , we do not find any `` natural '' explanation for the variability
luminosity relationship in the context of discrete - event models .
this is possibly an indication that explanations in which variability is not expressed in terms of discrete events should be favoured .
|
a very important assumption of the standard model of cosmology ( @xmath0 ) is based on the homogeneous and isotropic friedmann - lemaitre - robertson - walker solutions of einstein s equations . the homogeneity and the isotropy
are considered on large scale in the universe .
the universe is not isotropic or spatially homogeneus on local scales . the question of whether the universe is homogeneous and isotropic is of fundamental importance to cosmology , but we have not decisive answers . on the other hand
neither observations of luminosity distance combinated with galaxy number counts nor isotropic cosmic microwave background radiation are able to say if the universe is spatially homogeneous and isotropic .
the fundamental question consists in a simple observation : this geometry is the only that is able to explain and to be compatible with experimental data ?
are we sure that the assumption of homogeneity and isotropy is a logical and comforting way of thinking , or better is it an _ a - priori _ assumption ? + this is a pertinent question because we need that more @xmath1 of the content of our universe must be dark ( energy and matter ) in order to have a compatible model with observations .
the solution of the dark energy puzzle is the keystone of the modern cosmology . are there observables that can prove the universe is homogeneous and isotropic on large scales ?
very interesting studies has been done in this direction .
the @xmath2cdm model of the universe is remarkably successfull , but we have important tensions between the model and the experimental data . on the other hand dark energy is the biggest puzzle in cosmology .
there are many papers with more detailed discussions about dark energy , that are outside the scope of this paper , see for example @xcite and references therein .
there are many reasons that consider the @xmath2cdm model full of theoretical problems @xcite ,
one is that @xmath2 has a value absurdly small in quantum physics .
moreover we can not expect that dark energy will have in future locally observable effects .
+ + the cosmic microwave background ( cmb ) has high isotropy and this is considered as a strong evidence of the homogeneity and isotropy of the universe , that is to say the universe is well described by means of a friedmann - lemaitre - robertson - walker ( flrw ) model .
the main indication for this model is due to the ehlers , geren and sachs theorem ( egs ) @xcite in 1968 .
this theorem is due to an earlier paper of tauber and weinberg @xcite in 1961 . in egs theorem
we consider the observers in an expanding universe , dust universe measures isotropic cmb and this implies that flrw metric is valid and the cosmological principle is also valid .
this theorem is important because it permits to have the homogeneity and isotropy not from experimental measurements of the isotropy of the universe but from the cmb .
but as we will discuss later , cmb radiation have small anisotropies with @xmath3 of amplitude !
+ + as regards the homogeneity of the universe it is important to note that the mass density of the universe is not inhomogeneous on scales much smaller than the hubble radius , in other terms the homogeneity is not true at all orders but we can assume to be valid on distance greater that 100 mpc .
many papers indicate this feature , see for example @xcite ( and references therein ) , where the author indicates evidences that galaxy distribution is spatially inhomogenous for @xmath4 .
+ the strong interest in inhomogenous cosmological models , in particular the so called lemaitre - tolman - bondi ( ltb ) model @xcite ( for more details see @xcite and references therein ) , that represents a spherically symmetric exact solution to the einstein s equations with pressureless ideal fluid , is due to the its simplicity and it is very useful .
in fact it allows for studies of inhomogeneities that can not be analyzed as perturbative deviations from flrw and it permits to evaluate the effect of inhomogeneities in particular it has been studied that ltb models without dark energy can fit observed data .
+ the high precision cosmology is able to understand by more details our study about universe .
when we consider the isotropy of cmb we must not forget that this is not sufficient to say that our region of space is isotropic @xcite .
we have two very important observational evidences showing that we do nt have exact isotropy @xcite .
both evidences may be caused by an anisotropic phase during the evolution of our universe in other terms the existence of anomalies in cmb suggests the presence of an anomalous plane - mirroring symmetry on large scales .
the same anomalous features in seven - years wmap data and planck data seems to suggest that our universe could be non - isotropic .
the first is the presence of small anisotropy deviations as regards the isotropy of the cmb .
in fact we have small anisotropies with @xmath3 amplitude .
the second is connected with the presence of large angle anomalies @xcite .
these anomalies can be considered in 4 families : \1 ) the alignment of quadrupole and octupole moments ; \2 ) the large scale asymmetry ; \3 ) the very strange cold spot @xcite ; \4 ) the low quadrupole moment of the cmb , that is very important because it may indicate an ellipsoidal - bianchi type i anisotropic evolution of the universe .
this is due to the fact that the low quadrupole moment is suppressed at large scale and this suppression can not be explained by the common cosmological model . + some years ago it has been shown @xcite that if we start with a flrw universe , it is possible to have small deviations from homogeneity and isotropy taking into account small deviations in the cmb . in particular
if we consider an homogeneous and anisotropic universe , the small quadrupole anisotropy in cmb implies a very small anisotropy in the universe .
next , general results have been established @xcite , in which the authors does not assume a priori homogeneity and they found that small anisotropies in cmb imply that the cosmo is not exactly flrw but it is almost flrw .
limits on anisotropy and inhomogeneity can be found starting from cmb .
+ the cosmological model that takes into account all these and stimulates many interest is the `` anisotropic bianchi type i model '' that can be an intriguing alternative to the standard model flrw , in which small deviations from the isotropy is able to explain the anisotropies and the anomalies in the cmb .
+ the anisotropy considered in this work might be interpreted as an imprinting , a primordial relic of an early anisotropy that appears in the context of a multi- dimensional cosmological model of unified string theories . in this paper
our goals are to study an anisotropic and inhomogeneous model of the universe .
in particular we introduce a new approach to a universe in which inhomogeneities and anisotropies coexist , therefore we study in order to obtain the relative einstein s equations . these models of universe inhomogeneous and anisotropic has been studied in different physical situations , as the role of the diffusion forces in governing the large - scale dynamics of inhomogeneous and anisotropic universe @xcite .
the supernovae observations are good tests about the structure of the space - time on different scales .
this is a very important point , in fact some years ago zeldovich @xcite studied the importance of the effects of the inhomogeneities on light propagation and also in the next years @xcite . to check this model
we calculate the luminosity distance in order to compare theoretical approach with experimental data .
we explain the acceleration of the universe without invoking the presence of a cosmological constant or dark energy .
+ + the structure of this paper is the following . in the next section
we calculate the metric for this ltb - bianchi i model of the universe . in section iii after providing the calculation of various symbols we write the einstein s equations taking into account this geometry .
section iv is dedicated to calculate the luminosity distance and in section v we compare the theoretical data with experimental data .
finally the discussion and conclusion are summarized in section vi . + +
in order to find the anisotropic - ltb metric , let us start with a bianchi type i space - time metric , spatially homogeneous , descripted by the metric @xmath5 with two expansion parameters @xmath6 and @xmath7 that are the scale factors normalized in order that @xmath8 and @xmath9 present cosmic time .
the metric ( [ bianchi1 ] ) considers the xy - plane as a symmetry plane . to our aim
we write the bianchi type i metric in polar coordinate @xmath10 : @xmath11 \ , dr^2 -r^2 \ , [ a^2(t ) { \text{cos}}^2 \theta \nonumber \\ & & + b^2(t ) \ , { \text{sin}}^2 \theta ] \ , d \theta^2 -2r \ , [ a^2(t ) - b^2(t ) ] \text{sin } \theta \ , \text{cos } \theta \ , dr \ , d \theta \nonumber \\ & & - r^2 \ , a^2(t ) \ , { \text{sin}}^2 \theta \ , d \phi^2.\end{aligned}\ ] ] in order to have a ltb - bianchi i metric , we make the following substitutions @xmath12 @xmath13 in this way it is possible to obtain the general ltb - bianchi i metric in polar coordinate , observing that @xmath14 and @xmath15 where @xmath16 , we have : @xmath17 it is important to observe that the eq .
( [ ltb - bianchi1 ] ) brings back to known cases : @xmath18therefore the metric ( [ ltb - bianchi1 ] ) is a non - homogeneous metric with axial symmetry , that is simple referable to pure homogeneous or pure isotropic case .
+ let us define the following quantity @xmath19 that represents the degree of anisotropy of the universe . from the definition of @xmath20 we obtain : @xmath21
let us introduce these relations in the metric ( [ ltb - bianchi1 ] ) in order to show it as a function of @xmath20 and @xmath22 ( or @xmath20 and @xmath23 ) .
putting all togheter we have @xmath24dr^2+\nonumber\\&-&\left [ { a_\rvert}^2+(\epsilon^2 + 2\,a_\rvert\,\epsilon ) \sin^2 \theta \right ] d\theta^2+\nonumber\\&+&\left [ \left(\epsilon^2\right)'+2\,a'_\rvert\,\epsilon+2\,a_\rvert\,\epsilon ' \right]\sin \theta \cos \theta\,dr d\theta + \nonumber \\&-&{a_\rvert}^2 \sin^2 \theta d\phi^2\equiv\nonumber\\&\equiv & \left(g_{\rvert\mu \nu}^\text{(ltb)}+\delta g_{\rvert\mu \nu}^{(an)}\right)dx^\mu dx^\nu\end{aligned}\ ] ] with our metric given by @xmath25 where : @xmath26\\ \label{eq : pert113 } \delta g_{\rvert22}^{(an)}=&-\sin^2 \theta\,\left(\epsilon^2 + 2\,a_\rvert\,\epsilon\right).\end{aligned}\ ] ] the script `` ( ltb ) '' ( up or down is the same ) means that the quantity refers to lemaitre - tolman - bondi universe , while `` ( an ) '' refers to anisotropic universe . in other words
the metric ( [ eq : ltbplusanisotropic2 ] ) is be able to describe the inhomogeneity and axial anisotropy of the universe , on the other hand a very interesting thing is that it has been decomposed in the sum of a ltb metric with null curve and a a term that contains whole information about the anisotropy , @xmath27 .
+ for completeness reasons , it is possible to rewrite in the symmetric way the metric as @xmath28 where @xmath29 is obtained by eq .
( [ eq : pert10 ] ) with the substitution of @xmath23 instead of @xmath22 and @xmath30\\ \label{eq : pert213 } \delta g_{\perp 22}^{(an)}= & \cos^2 \theta \left ( \epsilon^2 - 2\,a_\perp\,\epsilon \right)\\ \label{eq : pert214 } \delta g_{\perp 33}^{(an)}= & \sin^2 \theta \left ( \epsilon^2 - 2\,a_\perp\,\epsilon \right).\end{aligned}\ ] ] + +
in this section we want to write the einstein s equations taking into account the ltb - bianchi i metric . to this end
we suppose a very small anisotropy of the universe , in order to have [ eq : limit_condition ] @xmath31 these positions permit to expand our results to the first order in @xmath20 , in other terms : @xmath32 with @xmath33 @xmath34 @xmath35 at this point we can calculate the christoffel connection to the first order in @xmath20 ( we repeats that the script @xmath36 and @xmath37 are indifferently written up or down ) : @xmath38 , % \end{aligned}\ ] ] that , neglecting the second order terms in @xmath39 , becomes @xmath40 the first term in eq .
( [ newgamma ] ) is just @xmath41 , the christoffel connection with the metric tensor @xmath42 , and putting @xmath43 and @xmath44 it is possibile to write the christoffel connection at the first order as @xmath45 let us calculate the ricci tensor @xmath46 with @xmath47 given by eq.([christoffelconnection ] ) .
therefore we have @xmath48 when we multiply in eq .
( [ ricci_explicate ] ) , neglecting the second order terms @xmath49 , @xmath50 , @xmath51 and @xmath52 , and putting @xmath53 @xmath54 @xmath55the ricci tensor to the first order in @xmath56 can be written as @xmath57 + in order to consider the perturbations of the energy momentum tensor , we consider a general anisotropic density energy given by : @xmath58 where @xmath59 .
the density that we choose has a planar symmetry , because of the consistency with the metric that we are working with .
this choice allows us to rewrite the energy momentum tensor and its trace as : @xmath60 @xmath61 where , in general @xmath62 .
% \]]however , all these definitions must be view as a first order correction to the usual energy momentum tensor in the ltb case .
this point of view becomes clear if we look at the usual perturbation theory provided by @xcite .
in fact , our particular definition is fully consistent with the mukhanov s one whether we fix @xmath63 in eq .
( 5.1 ) of @xcite : this means that we are only considering the perturbation in the energy density and we neglect the effect of a different pressure along two directions ( in particular , we continue in using a pressureless matter fluid everywhere ) . for sure , what we did is a strong constraint . by the way , the particular choice of the perturbation is not relevant for the purposes of this paper . + in order to be explicit we write : @xmath64 and @xmath65 as regards the energy conditions , we consider the general energy - momentum tensor : @xmath66 with @xmath67 . hence , energy conditions state : * weak energy condition : @xmath68 * dominant energy condition : by defining @xmath69 , @xmath70 * strong energy condition : @xmath71 * null energy condition : @xmath72 , where @xmath73 is a light - like vector .
a well - known , the hierarchy among these conditions is the following : strong implies null , dominant implies weak and weak implies null . in this way , by providing that the dominant condition holds , also weak and null are satisfied as well .
in particular , for energy - momentum ( 36 ) , with @xmath74 , they become : * weak : @xmath75 * dominant : @xmath76 * strong : @xmath77 * null : @xmath75 .
the einstein s equations in this universe are : @xmath78 .
\nonumber \\ & & \end{aligned}\ ] ]
the concept of distance depends on the assumed model of the universe and on the matter distribution in it .
the measured distance are influenced by inhomogeneities and anisotropy of the universe , see for example @xcite and @xcite .
the luminosity distance is one of the most important quantity to understand the presence of dark energy in the universe , considering the photon coming from supernovae ia . in this section
we want to calculate the luminosity distance for our metric eq.([eq : ltbplusanisotropic2 ] ) .
the reciprocity theorem by etherington ( 1993 ) @xcite and popolarized by ellis @xcite connects the angular diameter distance @xmath79 and the luminosity distance @xmath80 by @xmath81 where @xmath82 with @xmath83 temporal affine parameter and @xmath84 quadri - momentum of a generic signal that is started from the supernova and reaches us . to our end
it is necessary to calculate @xmath85 : @xmath86 with @xmath87 where we define @xmath88 in this way we obtain : @xmath89 this permits to write eq.([eq : anisotropicda1 ] ) as @xmath90 in the last equation we have considered the general relation between partial derivates in the coordinates @xmath91 and total derivates in the affine time @xmath83 .
in fact if @xmath92 is a generic function that depends from the coordinates it is possible to write : @xmath93 on the other hand as regards @xmath94 , we must write @xmath95 but we are considering radial signal , therefore @xmath96 , that is to say @xmath97 in this way we can consider @xmath98 as a parameter that is able to locate the trajectory of propagation of light .
this employment permits to write : @xmath99 therefore , eq . becomes : @xmath100 as regards the partial derivative of @xmath101 , remembering that we are considering the radial propagation of signals , the relevant components are : @xmath102 from which we have : @xmath103 in order to complete the analysis observe that @xmath104 and @xmath105 @xmath106 where we have put @xmath107 + now we work in small approximation of anisotropy , in order to use eq.([christoffelconnection ] ) to the lower order , in this way it is possibile to write : @xmath108 where @xmath109 .
therefore eq.([nabla ] ) is @xmath110 taking into account eq.([eq : usefullrelation ] ) it is possible to write the first two terms in eq.([eq : appr1 ] ) as @xmath111 . + inserting eq.([eq : appr1 ] ) in eq .
( [ dist2 ] ) it is possible to obtain @xmath79 , in fact we have : @xmath112 and integrating in @xmath83 we obtain : @xmath113 this expression of the luminosity distance reduces to the isotropic limit of the ltb metric , in fact we have @xmath114 therefore we obtain the limit @xmath115 + +
in this section we want to calculate the luminosity distance in order to obtain an operative expression and therefore to apply it to experimental data . the eq .
( [ eq : aniltbda ] ) give us the luminosity distance : @xmath116 but this expression is not directly applicable , because of it depends on @xmath117 coordinates and on the redshift .
theferore it is necessary to find the relations @xmath118 e @xmath119 .
to this end let us consider the definition of redshift and let us use the static observators classes that are geodetic also ( @xmath120 ) .
we consider light signal , that is to say @xmath121 , in this way we write : @xmath122 where @xmath123 e @xmath124 .
now we derive with respect to @xmath83 and we have @xmath125 on the other hand for geodetic radial signals we have @xmath126 e @xmath127 that gives @xmath128 as regards the ambiguity of the sign we must consider the minus sign because of increasing the distance ( @xmath129 ) we have a more ancient signal ( @xmath130 ) . when we consider the signals that respectively start at time @xmath131 and @xmath132 , the eq .
( [ eq : anilightpropagation ] ) must be valid , therefore we have : @xmath133 eq.([eq : anigeodesic2 ] ) can be written as @xmath134\frac{dr}{d\tau}\ ] ] that , taking into account eq.([eq : anigeodesic1 ] ) , can be written as @xmath135 now eqs .
( [ eq : dtdtau1 ] ) and ( [ eq : dtdtau2 ] ) are equal , therefore we have @xmath136 as regards @xmath119 it is important to remember that @xmath137 in this way taking into account eq .
( [ eq : anigeodesic2 ] ) we obtain the relation : @xmath138 putting all togheter , we are be able to write the luminosity distance as a function of the redshift @xmath139 and the angle @xmath98 @xmath140^\frac{1}{2 } \phantom{xxxxxxxxx}\\ \label{eq : rthetavsz } \frac{dr_\theta(z)}{dz}=&\frac{1}{1+z}\ ; \frac{1}{\partial_0 x(r_\theta(z),t_\theta(z),\theta)}\\ \label{eq : tthetavsz } \frac{dt_\theta(z)}{dz}=&-\frac{1}{1+z } \ ; \frac{x(r_\theta(z),t_\theta(z),\theta)}{\partial_0x(r_\theta(z),t_\theta(z),\theta)}.\end{aligned}\ ] ] it is important to observe that subscript @xmath98 remember us that the functions @xmath141 and @xmath131 are determined by the resolution of the system given by eqs ( [ eq : rthetavsz ] ) and ( [ eq : tthetavsz ] ) , where the angle @xmath98 is fixed and considered as a constant parameter during the propagation of light .
the accelerating expansion of the universe is driven by mysterious energy with negative pressure known as dark energy . in spite of all the observational evidences ,
the nature of dark energy is still a challenging problem in theoretical physics , therefore there has been a new interest in studying alternative cosmological models @xcite .
+ in the context of flrw models the acceleration of the universe requires the presence of a cosmological constant .
but it does not appear to be natural to introduce the presence of a cosmological constant and does not appear to be natural to introduce the dark energy .
+ in this section we consider the comparison between experimental data , in particular with union 2 data set of supernovae ia and our inhomogeneous and anisotropic universe .
let us suppose a small anisotropy in order to write the functions @xmath142 e @xmath143 as solutions of a ltb universe with null curvature and matter dominated .
we have @xmath144 where we have considered the following parametrization @xmath145 in this way we have the possibility to obtain again the simple model in which @xmath146 , @xmath147 and @xmath148 .
let us consider that today and in our position in the universe ( @xmath149 e @xmath150 ) the hubble constant is @xmath151 @xcite .
we have the following conditions @xmath152 , therefore we have : @xmath153 in this way we have not the six parameters of the model , now they are four : @xmath154 e @xmath155 . at this point
we remember the limits given by , therefore for @xmath156 we have @xmath157 this condition must be transfered to the four parameters .
+ + + the condition eq . is obtained when @xmath158 and @xmath159 . on the other hand
it also must be @xmath160 .
therefore we have : @xmath161 from which we obtain @xmath162 @xmath163 in this way , for @xmath156 and @xmath160 , we must write : @xmath164 therefore @xmath165 , from which we obtain @xmath166 . in conclusion
we have also three parameters : @xmath167 and @xmath168 .
the advantage of this parametrization is that we can change @xmath169 and @xmath170 as we want , taking into account that @xmath171 . in our work
we have changed @xmath168 in the range @xmath172 $ ] . [ cols="^,^,^",options="header " , ] in table
we have the best fit values of the parameters for @xmath173 . in fig .
1 we have the hubble diagram for the 557 supernovae ia of the union 2 catalog .
the best fit curve is in the same diagram .
the fit of the cosmological observational data is in very good agreement , without using any dark energy ! according to our ansatz , it is important to stress that @xmath174 , so dominant energy condition up to first order gives : @xmath175 while the other ones give : @xmath176 + furthermore , @xmath177 so , from the einstein s equation , we have that : @xmath178\nonumber\\ \delta_{mat}&=\frac{1}{8\pi\,g}\left [ \left(\frac{\dot a_\perp}{a_\perp}\right)^2 + 2\frac{\dot a_\perp}{a_\perp } \frac{\dot a'_\perp}{a'_\perp}-\left(\frac{\dot a_\rvert}{a_\rvert}\right)^2 - 2\frac{\dot a_\rvert}{a_\rvert } \frac{\dot a'_\rvert}{a'_\rvert } \right].\end{aligned}\ ] ] + hence we have from solutions ( 70a ) and ( 70b ) , with conditions ( 75a ) , ( 75b ) and ( 77 ) , that eq . gives a constraint on parameters which must be satisfied .
in particular , by using the best - fit values for @xmath170 and @xmath169 , dominant energy condition requires @xmath179 , which is in fully agreement with our analysis .
in the present paper we have studied the possible effects of an anisotropy and inhomogeneity in the expansion of the universe .
+ the motivation behind this choise is that singly , inhomogeneous cosmological models and bianchi i cosmological model of the universe have motivations of thruth that must not be left out for one of the the two models .
both models may be unified in a anisotropic expansion of the inhomogeneous universe .
the ltb - bianchi i model posseses important specific properties and at the same time this is not too complicated from a physical and mathematical point of view .
+ in particular we have connected this model on present - day observations as luminosity distance of the supernovae ia .
we fit observational data from union 2 catalog of the supernovae ia with a ltb - bianchi i model of the universe .
the agreement is good .
we have not any dark energy in this model . + + we are sure that the voids in the universe dominate , while matter is distributed in a filamentary structure .
therefore photons must travel through the voids and the presence of inhomogeneities can alter the observable with respect to the corresponding flrw model of universe , homogeneous and isotropic .
+ the key point is that in this model we have two contributions to the hubble diagram of the supernovae ia : inhomogeneity to the large scale geometry and anisotropy can generate dynamically effects that may remove the need for the postulate of dark energy .
+ this model must be intended as a first step towards a most general case .
the model is oversimplifying for different reasons .
first , we have considered only the first order in @xmath20 .
second , it is necessary to generalize this paper , a very interesting open question that we will study in future , is to obtain how to treat light - cone average in more realistic cosmological calculation .
third we have considered he simple ltb model of the universe , but may be very interesting to study more completed inhomogeneous model as for example swiss - cheese model .
note , finally that several possibility are allowed by our model , it will be interesting to compare this model with other experimental cosmological data .
+ in future we will study the possibility that inhomogeneity and anisotropy can have significant effects on the propagation of light , with potentially very important effects on cosmological observations and we want to study different observational tests that may confirm this model of the universe
. + + + + the authors would like to thank m. gasperini for useful discussions .
this work is supported by the research grant `` theoretical astroparticle physics '' number 2012cppyp7 under the program prin 2012 funded by the ministero dellistruzione , universit e della ricerca ( miur ) .
99 j. ehlers , p. geren and r. k. sachs , j. math .
* 9 * , 1344 ( 1968 ) .
c. a. clarkson and r. barrett , class .
* 16 * , 3781 ( 1999 ) . c. clarkson and r. maartens , class .
* 27 * , 124008 ( 2010 ) .
w. percival , mon not .
. soc . * 381 * , 1053 ( 2010 ) .
h.j . de vega , m.c .
falvella and n.g .
sanchez ( 2010 ) arxiv:1009.3494 .
a. blanchard , astron .
astrophys . rev . * 18 * , 595 ( 2010 ) .
t. clifton , p.g .
ferreira , a. padilla and c. skordis ( 2011 ) arxiv:1106.2476 .
copeland , m. sami and s. tsujikawa , int .
d * 15 * , 1753 ( 2006 ) .
g. e. tauber and j. w. weinberg , phys .
* 122 * , no . 4 , 1342 ( 1961 ) .
f. s. labini , class .
* 28 * , 164003 ( 2011 ) .
g. lemaitre , annal soc .
sci bruxelles * 53 * , 51 ( 1933 ) ; gen rel .
grav . * 29 * , 641 ( 1997 ) .
tolman , proc . national acad
. sci . * 20 * , 169 ( 1934 ) .
h. bondi , mon . not .
. soc . * 107 * , 410 ( 1947 ) .
see for example a. krasinsky , _ inhomogeneous cosmological models _
, cambridge university press ( 2006 ) .
r. maartens , phil .
a * 369 * , 5155 ( 2011 ) .
e. russel , c.b .
kilinc and o. pashev , arxiv:1312.3502 .
v.g . gurzadyan _
_ , a&a * 490 * , 929 ( 2008 ) .
v.g . gurzadyan _
et al_. , a&a * 498 * , l1 ( 2009 ) .
bennet , r.s .
hill , g. hinshaw , m.r .
nolta , n. odegard , l. page , d.n .
spergel , j.l .
weiland , ap.js * 148 * , 97 ( 2003 ) .
k. land and j. magueijo , phys .
lett . * 95 * , 071301 ( 2005 ) .
ralston and p. jain , int .
j. mod .
d * 13 * , 1857 ( 2004 ) .
a. de oliveira - costa , m. tegmark , m. zaldarriaga and a. hamilton , phys .
d * 69 * , 063516 ( 2004 ) . c. j. copi , d. huterer and g. d. starkman , phys .
d * 70 * , 043515 ( 2004 ) .
h. k. eriksen , f. k. hansen , a. j. banday , k. m. gorski and p. b. lilje , astrophys .
j. * 605 * , 14 ( 2004 ) [ erratum - ibid .
* 609 * , 1198 ( 2004 ) ] .
f. k. hansen , a. j. banday and k. m. gorski , mon . not .
354 * , 641 ( 2004 ) .
p. vielva , e. martinez - gonzalez , r. b. barreiro , j. l. sanz and l. cayon , astrophys .
j. * 609 * , 22 ( 2004 ) .
barrow phys . rev .
d * 55 * , 7451 ( 1997 ) .
a. berera , r.v .
buniy and t.w .
kephart , j. cosmolog .
. phys . * 0410 * , 016 ( 2004 ) .
l. campanelli , p. cea and l. tedesco , phys .
lett . * 97 * , 131302 ( 2006 ) ; [ erratum-_ibid_. * 97 * 131302 ( 2006 ) .
l. campanelli , p. cea and l. tedesco , phys .
d * 76 * 063007 ( 2007 ) .
e. martinez - gonzalez , j.l .
sanz , a & a * 300 * , 346 ( 1995 ) .
r. maartens , g.f.r .
ellis , w. stoeger , phys .
d * 51 * , 1525 ( 1995 ) .
r. maartens , g.f.r .
ellis , w. stoeger , phys .
d * 51 * , 5942 ( 1995 ) .
d. shogin and s. hervik , jcap * 1310 * , 005 ( 2013 ) .
zeldovich , sov . astron .
aj * 8 * , 13 ( 1964 ) .
b. bertotti , proc .
a*294 * , 195 ( 1966 ) .
v.m . dashevskii and v.i .
slysh , sov .
astron.*9 * , 671 ( 1966 ) e.j .
gunn , astrophys .
j. * 150 * , 737 ( 1967 ) .
r. kantowski , astrophys .
j. * 155 * , 89 ( 1969 ) .
dyer and r. roeder , astrophys .
j. * 174 * , l115 ( 1972 ) .
s. weinberg , astrophys .
j. * 208 * l1 ( 1976 ) .
ellis , b.a .
bassett and p.k.s .
dunsby , class.quantum grav . * 15 * , 2345 ( 1998 ) . j. kristian and r.k .
sachs , astrophys .
j. * 143 * , 379 ( 1966 ) .
j. adamek , e. di dio , r. durrer and m. kunz , phys .
d * 89 * , 063543 ( 2014 ) .
etheringotn , philosophical magazine ser .
7 , vol 15 , 761 ( 1933 ) .
g. ellis , gen .
. grav . * 39 * 1047 ( 2007 ) .
v. f. mukhanov , h. a. feldman and r. h. brandenberger , phys .
* 215 * ( 1992 ) 203 .
[ planck collaboration ] astron . atsrophys .
|
in this paper we introduce a ltb - bianchi i ( plane symmetric ) model of universe .
we study and solve einstein field equations .
we investigate the effects of such model of universe in particular these results are important in understanding the effect of the combined presence of an inhomogeneous and anisotropic universe .
the observational magnitude - redshift data deviated from union 2 catalog has been analyzed in the framework of this ltb - anisotropic universe and the fit has been achieved without the inclusion of any dark energy .
|
the understanding of few - body systems relies in our capability to design methods for finding an exact or approximate solution of the @xmath0body problem . in two- , three- , and four - body problems it is possible to obtain mathematically correct and computationally tractable equations such as the lippmann - schwinger , faddeev and yakubovsky equations describing exactly , for any assumed interaction between the particles , the motion of few - body systems @xcite .
however , the exact solution of these equations requires sophisticated techniques whose difficulty increases when increasing the number of particles .
there are a countless number of examples of quantum - mechanical few - body systems , from few - electron quantum dots in solid state physics to constituent quarks in subnuclear physics .
the intricate feature of the few - body systems is that they develop individual characters depending on the number of constituent particles .
the most important cause of these differences are the correlated motion and the pauli principle .
this individuality requires specific methods for the solution of the few - body schrdinger equation , approximate solutions assuming restricted model spaces ( as for example the mean field approximation ) failing to describe these systems .
the solution of any few - particle system may be found in a simple and unified approach .
a recent widely used method is the stochastic variational @xcite , a variational approach where the trial wave function is generated by a random search on an adequate basis .
an alternative powerful tool is an expansion of the trial wave function in terms of hyperspherical harmonic ( hh ) functions .
the idea is to generalize the simplicity of the spherical harmonic expansion for the angular functions of a single particle motion to a system of particles by introducing a global length @xmath5 , called the hyperradius , and a set of angles , @xmath6 .
for the hh expansion method to be practical , the evaluation of the potential energy matrix elements must be feasible .
the main difficulty of this method is to construct hh functions of proper symmetry for a system of identical particles .
this is a difficult problem that may be overcome by means of the hh formalism based on the symmetrization of the @xmath0body wave function with respect to the symmetric group using the barnea and novoselsky algorithm @xcite .
this method , applied in nuclear physics for @xmath7 @xcite , has only been applied to quark physics for @xmath8 @xcite .
therefore , its generalization would be ideally suited for the study of the properties of multiquark systems . during the last few years
there has been a renewed interest on the possible existence of multiquark states , specially four- ( two quarks and two antiquarks ) and five - quark ( four quarks and one antiquark ) , in the low - energy hadron spectroscopy .
theoretically , the possible existence of four - quark bound states was already suggested thirty years ago , both in the light - quark sector by jaffe @xcite and in the heavy - quark sector by iwasaki @xcite .
experimentally , there were several analysis suggesting the existence of non-@xmath9 states in the high energy part of the charmonium spectrum @xcite .
the recent series of discoveries of new meson resonances whose properties do not fit into the predictions of the naive quark model , has reopened the interest on the possible role played by non-@xmath9 configurations in the hadron spectra @xcite . among them
one could mention the @xmath10 @xcite , the @xmath11 @xcite or the new @xmath12 and @xmath13 resonances @xcite .
out of the several interpretations proposed for these states , those based on four - quark states are majority @xcite
. there seems to be a consensus that a four - quark system containing two - light and two - heavy quarks , @xmath14 , is stable against dissociation into two mesons , @xmath15 , if the ratio of the mass of the heavy to the light quark is large enough @xcite .
while the conclusions agree for the @xmath16 system , there are more discrepancies about the @xmath17 system @xcite .
the possible existence of four - quark states in the low - lying meson spectra has been used to advance in the understanding of the light - scalar mesons , @xmath18 . in this case , four - quark states have been justified to coexist with @xmath9 states in the energy region below 2 gev because they can couple to @xmath19 without orbital excitation @xcite , and therefore conventional @xmath20 states are expected to mix with four - quark states to yield physical mesons @xcite .
the four - heavy quark states have not received as much attention in the last years as the ones containing light quarks .
iwasaki @xcite , based on the constituent quark mass value obtained within a string model for hadrons , argued that four - charm quark states could exist in the 6 gev energy region . using a variational method with gaussian trial radial wave functions and two - body potentials based on the exchange of color octets between quarks , ader _ et al . _
@xcite obtained non stable @xmath21 states . however , using a potential derived from the mit bag model in the born - oppenheimer approximation the same authors concluded that the @xmath22 state was bound by 35 mev .
heller and tjon @xcite considered also the mit bag model improving the born - oppenheimer approximation with a correct treatment of the kinetic energy , non stable @xmath21 states were found . in ref .
@xcite @xmath23 @xmath21 states were analyzed in the framework of a chromomagnetic model where only a constant hyperfine potential , not depending on radial coordinates , was retained . with these assumptions no
bound @xmath24 states were found .
a similar conclusion was obtained in ref .
@xcite using the bhaduri potential and solving variationally the four - body problem in a harmonic oscillator basis .
recently , lloyd and vary @xcite investigated the @xmath22 system using a nonrelativistic hamiltonian inspired by the one - gluon exchange potential diagonalizing the hamiltonian in a harmonic oscillator basis , obtaining several close - lying bound states .
the discussion above illustrates that the theoretical predictions for the existence of four - quark systems differ depending basically on the method used to solve the four - body problem and the interaction employed .
it is our aim in this work to make a general study of four - quark systems of identical flavor in an exact way . for this purpose
we will generalize the hh method @xcite , widely used in traditional nuclear physics for the study of few - body nuclei , to study four - quark systems .
there are two main difficulties , first the simultaneous treatment of particles and antiparticles , and second the additional color degree of freedom . as a test of our formalism
we will recover some results present in the literature .
finally , we will make a general study of four - quark systems of identical flavor by means of the constituent quark model of ref .
@xcite that provides with a realistic framework describing in a correct way the general aspects of the meson and baryon spectra .
the paper is organized as follows . in the next section
the formalism necessary to build a color singlet wave function with well - defined parity and @xmath25parity quantum numbers is described and discussed . in sect .
[ model ] we introduce the general features of the constituent quark model used . in sect .
[ results ] we present and analyze the results obtained for the four - quark systems . finally , in sect .
[ summary ] we resume our most important conclusions .
the system of two quarks and two antiquarks with the same flavor can be regarded as a system of four identical particles .
each particle carries a @xmath26 spin label and a @xmath27 color label .
both quarks and antiquarks are spin @xmath28 particles , but whereas a quark color state belongs to the @xmath27 fundamental representation @xmath29 $ ] , an antiquark color state is a member of the fundamental representation @xmath30 $ ] .
the four - body wave function is a sum of outer products of color , spin and configuration terms @xmath31 coupled to yield an antisymmetric wave function with a set of quantum numbers that reflects the symmetries of the system .
these are the total angular momentum quantum number @xmath32 , its projection @xmath33 , and the @xmath27 color state @xmath34 ( to avoid confusion with the charge conjugation quantum number the color degree of freedom will be labeled as @xmath34 from now on ) , which by assumption must belong for physical states to the @xmath27 color singlet representation .
since qcd preserves parity , parity is also a good quantum number . another relevant quantum number to the system under consideration it is the @xmath25parity , @xmath35 , i.e. , the symmetry under interchange of quarks and antiquarks . to obtain a solution of the four - body schrdinger equation we eliminate the center of mass and use the relative , jacobi , coordinates @xmath36 .
then we expand the spatial part of the wave - function using the hh basis .
in this formalism the jacobi coordinates are replaced by one radial coordinate , the hyperradius @xmath5 , and a set of ( @xmath37 ) angular coordinates @xmath38
. the hh basis functions are eigenfunctions of the hyperspherical part of the laplace operator .
an antisymmetric @xmath39body basis functions with total angular momentum @xmath40 , color @xmath41 and @xmath35-parity @xmath35 , are given by , @xmath42^{j_a j^z_a } | n { \rangle}\,,\end{aligned}\ ] ] where @xmath43 are the hyperradial basis functions , taken to be laguerre functions .
@xmath44}_{k_a l_a m_a \gamma_a y_{a-1 } \alpha_a}(\omega_a)\ ] ] are hh functions with hyperspherical angular momentum @xmath45 , and orbital angular momentum quantum numbers ( @xmath46 ) that belong to well - defined irreducible representations ( irreps ) @xmath47 of the permutation group subgroup chain @xmath48 , denoted by the yamanouchi symbol @xmath49 \equiv [ \gamma_a,\gamma_{a-1},\ldots,\gamma_1 ] $ ] .
the dimension of the irrep @xmath50 is denoted by @xmath51 and @xmath52 is a phase factor @xcite .
similarly , the functions @xmath53}_{s_a s^z_a g_a \ ,
\widetilde{\gamma}_{a},\widetilde{y}_{a-1 } \ , \beta_a}(s^z_1 .. s^z_a , g_1 .. g_a)\ ] ] are the symmetrized color
spin basis functions , given in terms of the spin projections ( @xmath54 ) and color states ( @xmath55 ) of the particles .
the quantum numbers @xmath56 , and @xmath57 are used to remove the degeneracy of the hh and color spin states , respectively . for the construction of the symmetrized hh basis we shall use the algorithm of barnea and novoselsky @xcite , which utilizes the group of kinematic rotations . for the color
spin subspace , we will present in the following subsections a method to transform the standard basis into a symmetrized color spin basis with well defined color and @xmath25parity .
the calculation of the hamiltonian matrix - elements between the antisymmetric basis functions , eq .
( [ hh_a ] ) , is practically the same as in the nuclear physics case , replacing isospin by color , and the reader is referred to barnea _ et .
_ @xcite .
the starting point for the construction of color - spin states with well - defined permutational symmetry , color and spin quantum numbers , are the single particle states .
the spin state of the @xmath58th particle , quark or antiquark , is given by @xmath59 .
using the clebsh - gordan coefficients one can successively couple the spin states to construct an @xmath39-body state with well defined total spin @xmath60^{s_2 } |{{\textstyle \frac{1}{2}}}{\rangle}]^{s_3 } \ldots |{{\textstyle \frac{1}{2}}}{\rangle}]^{s_a s^z_a}\,.\end{aligned}\ ] ] theoretically one can adopt the same procedure for the color states replacing the @xmath26 clebsh - gordan coefficients with the appropriate @xmath27 ones . in practice , however , the @xmath27 coefficients are more involved and we shall therefore follow a different path while dealing with color . each quark ( antiquark )
color state is one of the three states of the triplet representation @xmath29 $ ] ( @xmath30 $ ] ) of @xmath27 .
these states are eigenstates of the operators ( @xmath61 ) which form the cartan subalgebra of @xmath27 .
let us denote by @xmath62 the color state of the @xmath58th particle , @xmath63 for quark states and @xmath64 for antiquark states ,
@xmath65 the eigenvalues @xmath66 ( weights ) uniquely label the quark ( antiquark ) states , in particular we can write @xmath67 an @xmath39-body color state is a product of the single particle color states .
this state does not belong to a well - defined irreducible representation of the color @xmath27 group , but instead it is characterized by the `` color projections '' @xmath68 and @xmath69 , very much the same way as the `` m - scheme '' states used in nuclear structure calculations which have @xmath33 as a good quantum number but not @xmath32 . from now on we shall use the notation @xmath70 for the pair of eigenvalues @xmath71 .
the combined color - spin single particle state is @xmath72 , and the @xmath73-particle states are given by @xmath74^{s_2 } | { { \textstyle \frac{1}{2 } } } g_3{\rangle}]^{s_3 } \ldots | { { \textstyle \frac{1}{2 } } } g_n{\rangle}]^{s_n s^z_n } \ , .\ ] ] in the following we shall present a method to transform these states into basis states with well - defined permutational symmetry , color and @xmath25parity .
this method is based on an algorithm by novoselsky , katriel and gilmore ( nkg ) @xcite who devised a recursive method to split an invariant vector space into states with well - defined permutational symmetry . by now , this algorithm has been used to construct symmetrized states in broad range of problems , and therefore we shall not give a detailed account of the method but rather outline the main ideas . for a clear description of the method the interested reader
is referred to @xcite .
the main tool of the nkg algorithm is the transposition class - sum operator . for the @xmath73-particle permutation group @xmath75 ,
this operator is the sum over all possible transpositions , @xmath76=\sum_{i <
j}^n ( i , j ) \;.\ ] ] it turns out that a simultaneous eigenstate of the mutually commuting transposition class - sum operator set @xmath77 , { \ensuremath{\boldsymbol{\cal c}}}_2[{\cal s}_{n-1}],\ldots , { \ensuremath{\boldsymbol{\cal c}}}_2[{\cal s}_2]\}$ ] has definite symmetry with respect to the group - subgroup chain @xmath78 . accordingly , nkg formulated a recursive method that starts with one - particle states and constructs symmetrized @xmath39-particle states through the following procedure : * add another particle ( let say a total of @xmath73 ) . *
identify the subspaces invariant under the action of the permutation group @xmath75 .
create the appropriate basis .
* in each subspace , evaluate the matrix elements of the transposition class - sum operator @xmath79 $ ] , and diagonalize it . * the @xmath75 symmetry of the resulting states
is unambiguously identified through the eigenvalues of @xmath79 $ ] .
the eigenvectors of the @xmath79 $ ] matrix are the transformation coefficients from the old basis ( c. ) to the new symmetrized one , these are the coefficients of fractional parentage . * ensure phase consistency ( see reference @xcite for a detailed account of this delicate point ) . * if @xmath80 repeat the procedure .
when this procedure is done , the original states are transformed step by step into the desired symmetrized @xmath39-particle states . in the case under consideration ,
the @xmath73-particle states are labeled by a set of good quantum numbers consisting of the total spin @xmath81 and its projection @xmath82 ( which we shall suppress ) , the color projection @xmath83 and the set of young diagrams @xmath84 which is equivalent to the yamanouchi symbol @xmath85 @xcite .
using these labels one can construct a complete set of states labeled by @xmath86 , where @xmath87 is an additional label that removes the remaining degeneracies .
for single particle , there is no degeneracy and one sees that @xmath88 where @xmath89 and @xmath90 or equivalently @xmath91 . to obtain a two - particle state labeled by the sequence of young diagrams @xmath92 , we first couple the spin states , and introduce the notation @xmath93^{s_2}\ ] ] to denote @xmath94-particle states with total spin @xmath95 and total color projection @xmath96 . here , and
in what follows , we have used the short hand notation @xmath96 for @xmath97 .
the permutation operators transform these states within an invariant subspace , each of which is characterized by the total spin @xmath95 and the total color projection @xmath98 .
the linear combinations which belong to well - defined irreps of the symmetry group can be written in the form , @xmath99 | ( s_1 g_1 \gamma_1 ; g_2 ) s_2 g_2 { \rangle}\;.\ ] ] assume that we have constructed a @xmath100particle states with well - defined permutational symmetry . following the @xmath101particle example we first form @xmath73-particle states with well - defined total spin and total color projection , @xmath102^{s_n } \,.\ ] ] note that @xmath103 .
the desired , symmetrized , @xmath73-particle states can now be written in the form @xmath104 the transformation coefficients @xmath105 $ ] are the color - spin coefficients of fractional parentage , cscfps .
these coefficients are the lines of an orthogonal transformation matrix .
therefore they fulfill the orthogonality and completeness relations .
the evaluation of the matrix - elements of the transposition class - sum operator , @xmath79 $ ] , for the non - symmetrized @xmath73-particle states , eq .
( [ n_nonsym ] ) , can be drastically simplified if we recast it in the form @xmath106={\ensuremath{\boldsymbol{\cal c}}}_2[{\cal s}_{n-1 } ] + \sum_{i=1}^{n-1 } ( i , n ) \,.\ ] ] the first term on the right - hand side of eq .
( [ c2sr ] ) is diagonal in @xmath107 .
its explicit value can be calculated exploiting the following expression for the eigenvalues of @xmath79 $ ] @xcite , @xmath108=\frac{1}{2}\sum_{i } r_i(r_i-2i+1 ) \
, , \ ] ] where @xmath109 is the number of boxes in the @xmath58th row of @xmath110 and the sum extends over all the rows .
the matrix elements of the second term between states with well - defined @xmath111 symmetry can be rewritten as @xcite , @xmath112 the @xmath113 matrix elements can be evaluated by expanding the @xmath100particle states in terms of the cscfps calculated previously and by recoupling the spin states in different order , @xmath114 \cr & & [ s_{n-2 } g_{n-2 } y_{n-2 } \beta_{n-2 } g'_{n-1 } |\ } s'_{n-1 } g'_{n-1 } y_{n-1 } \beta'_{n-1 } ] \cr & & \times \sum_{s_{n , n-1 } } { \left\ { \begin{array}{ccc } { { \textstyle \frac{1}{2 } } } & { { \textstyle \frac{1}{2 } } } & s_{n , n-1 } \\ s_{n-2 } & s_n & s_{n-1 } \end{array } \right\ } } { \left\ { \begin{array}{ccc } { { \textstyle \frac{1}{2 } } } & { { \textstyle \frac{1}{2 } } } & s_{n , n-1 } \\ s_{n-2 } & s_n & s'_{n-1 } \end{array } \right\ } } \cr & & ( -1)^{s_n+s_{n-2 } } ( 2 s_{n , n-1}+1 ) \sqrt{(2 s_{n-1}+1)(2 s'_{n-1}+1 ) } \cr & & { \langle}s_{n , n-1 } g_{n-1 } g_n |(n-1,n)| s_{n , n-1 } g'_{n-1 } g'_n { \rangle}\,.\end{aligned}\ ] ] since the transposition @xmath113 commutes with the @xmath115particle operators the matrix element is diagonal in the @xmath115particle states .
the matrix element on the right - hand side of eq .
( [ nnm1 ] ) is the matrix element between two - particle states after we have decoupled them from the rest of the system .
this matrix element can easily be evaluated to yield , @xmath116 in the previous subsection we outlined a method to construct color
spin states with well defined permutational symmetry @xmath117 , spin quantum numbers @xmath118 , and color projection @xmath70 . still , the @xmath27 color symmetry and @xmath25parity of these states is not established , since a state with the same color projection can belong to numerous irreps of @xmath27 .
for example the color projection @xmath119 can belong to the singlet representation @xmath120 $ ] as well as to the octet representation , @xmath121 $ ] .
the @xmath27 symmetry of the @xmath39-particle states can be established using the quadratic casimir operator , @xmath122 $ ] .
any @xmath27 state is an eigenvector of @xmath122 $ ] with eigenvalue that depends on the specific representation of the state . in particular the states with eigenvalue @xmath123 are singlet @xmath27 states .
the @xmath39-particle @xmath27 generators @xmath124 are symmetric with respect to particle permutations .
consequently the quadratic casimir , @xmath125={{\textstyle \frac{1}{4 } } } \vec{{\lambda}}^c \cdot \vec{{\lambda}}^c = { { \textstyle \frac{1}{4}}}\sum_{a=1}^{8 } { \lambda}^c_a { \lambda}^c_a \;,\ ] ] commutes with the permutation group @xmath126 , and the states @xmath127 with good quantum numbers @xmath128 form an invariant subspace for @xmath122 $ ] .
the calculation of the matrix - elements of the quadratic casimir can be simplified if we rewrite it as a sum of one - particle and two - particle terms @xmath129 = { { \textstyle \frac{1}{4 } } } \sum_{i=1}^a \vec{{\lambda}}^c_i \cdot \vec{{\lambda}}^c_i + { { \textstyle \frac{1}{4 } } } \sum_{i\neq j}^a \vec{{\lambda}}^c_i \cdot \vec{{\lambda}}^c_j \,.\ ] ] the first sum on the right - hand side of eq .
( [ c2_su3_a ] ) is just a sum of the single particle quadratic casimir operator , and its matrix - elements are just latexmath:[\[{\langle}s_a g_a y_a \beta_a | { { \textstyle \frac{1}{4 } } } \sum_{i=1}^a \vec{{\lambda}}^c_i \cdot \vec{{\lambda}}^c_i since the eigenvalue of the quadratic casimir for both representations @xmath29 $ ] and @xmath30 $ ] equals @xmath131 . due to schur s lemma the matrix elements of @xmath122
$ ] are independent of the particular permutational symmetry state @xmath132 . exploiting this observation we can recast the second sum on the right - hand side of eq .
( [ c2_su3_a ] ) into the form @xmath133 the @xmath134 matrix elements can be evaluated by expanding the @xmath39-particle states in terms of the cscfps , eq .
( [ cscfps_n ] ) , to decouple the last two particles , @xmath135 [ s'_{a-1 } g'_{a-1 } y_{a-1 } \beta'_{a-1 } g'_a |\ } s_a g_a y_{a } \beta'_a ] \cr & & \times [ s_{a-2 } g_{a-2 } y_{a-2 } \beta_{a-2 } g_{a-1 } |\ } s_{a-1 } g_{a-1 } y_{{a-1 } } \beta_{a-1 } ] \cr & & \times [ s_{a-2 } g_{a-2 } y_{a-2 } \beta_{a-2 } g'_{a-1 } |\ } s_{a-1 } g'_{a-1 } y_{{a-1 } } \beta'_{a-1 } ] \cr & & \times { \langle}g_a g_{a-1 } | \vec{{\lambda}}^c_{a-1 } \cdot \vec{{\lambda}}^c_a | g'_a g'_{a-1 } { \rangle}\;.\end{aligned}\ ] ] the last term on the right - hand side of eq .
( [ c2b ] ) is the sum @xmath136 which can be easily evaluated using the explicit form of the @xmath27 generators for the @xmath29 $ ] and @xmath30 $ ] irreps , eq .
( [ su3_33bar_states ] ) .
turning now to establish states with good @xmath25parity , we note that the @xmath25parity operator , @xmath35 , commutes with @xmath126 , the spin operator and the quadratic casimir of @xmath27 .
therefore the states @xmath127 with good quantum numbers @xmath128 that belong to the singlet irrep of @xmath27 form an invariant subspace for @xmath35 . in order to evaluate the matrix - elements of @xmath35
we first express our states in terms of the spin coupled single - particle states , eq .
( [ chi_cs ] ) .
then we can replace each quark state by the appropriate antiquark state and vice versa .
the eigenvectors of @xmath35 with eigenvalues @xmath137 that belong to the subspace of @xmath122 $ ] eigenstates with eigenvalue 0 are the desired physical color - spin basis functions .
the charmonium spectra has been thoroughly studied using constituent quark models for more than 25 years . as an almost pure coulombic system
, models based on coulomb plus confinement interactions have been used to describe most of its spectroscopic properties @xcite . to address the study of four - quark states we make use of a standard constituent quark model inspired in these works and originally applied to the study of the nonstrange baryon spectra and the baryon - baryon interaction @xcite .
this model has been generalized to all flavor sectors giving a reasonable description of the meson spectra @xcite , the baryon spectra @xcite , and the scalar mesons once four - quark configurations were included @xcite . within this model hadrons
are described as clusters of constituent ( massive ) quarks .
this is based on the assumption that the light - quark constituent quark mass appears because of the spontaneous breaking of the original @xmath138 chiral symmetry at some momentum scale . in this domain of momenta , quarks interact through goldstone boson exchange potentials . for the particular case of heavy quarks ,
chiral symmetry is explicitly broken and therefore the boson exchanges do not contribute to the @xmath139 interaction .
explicit expression of the boson exchange interacting potentials and a more detailed discussion can be found in ref .
@xcite . beyond the chiral symmetry breaking scale
one expects the dynamics being governed by qcd perturbative effects .
they are taken into account through the one - gluon - exchange ( oge ) potential @xcite .
the nonrelativistic reduction of the oge diagram in qcd for point - like quarks presents a contact term that , when not treated perturbatively , leads to collapse @xcite .
this is why one maintains the structure of the oge , but the @xmath140 function is regularized in a suitable way .
this regularization is justified based on the finite size of the constituent quarks and should be therefore flavor dependent @xcite .
as a consequence , the oge reads , @xmath141 where @xmath142 are the @xmath27 color matrices , @xmath143 is the quark - gluon coupling constant , and @xmath144 , where @xmath145 is the reduced mass of the interacting quarks @xmath146 ( @xmath73 stands for the light quarks @xmath147 and @xmath148 ) and @xmath149 is a parameter to be determined from the data .
the strong coupling constant , taken to be constant for each flavor sector , has to be scale - dependent when describing different flavor sectors @xcite .
such an effective scale dependence has been related to the typical momentum scale of each flavor sector and assimilated to the reduced mass of the system @xcite .
we use a strong coupling constant given by , @xmath150 where @xmath151 , @xmath152 and @xmath153 are parameters fitted within a global description of the meson spectra @xcite .
finally , any model imitating qcd should incorporate confinement .
lattice calculations in the quenched approximation derived , for heavy quarks , a confining interaction linearly dependent on the interquark distance .
the consideration of sea quarks apart from valence quarks ( unquenched approximation ) suggest a screening effect on the potential when increasing the interquark distance @xcite .
creation of light - quark pairs out of vacuum in between the quarks becomes energetically preferable resulting in a complete screening of quark color charges at large distances .
string breaking has been definitively confirmed through lattice calculations @xcite in coincidence with the quite rapid crossover from a linear rising to a flat potential well established in su(2 ) yang - mills theories @xcite .
a screened potential simulating these results can be written as @xmath154 at short distances this potential presents a linear behavior with an effective confinement strength @xmath155 , while it becomes constant at large distances .
such screened confining potentials provide with an explanation to the missing state problem in the baryon spectra @xcite , improve the description of the heavy meson spectra @xcite , and justify the deviation of the meson regge trajectories from the linear behavior for higher angular momentum states @xcite .
the results for the @xmath22 states have been obtained in the framework of the hyperspherical harmonic formalism explained above up to the maximum value of @xmath156 within our computational capabilities , being @xmath157 for positive parity states and @xmath158 for negative parity states with the exception of @xmath159 ( @xmath160 ) , @xmath2 and @xmath3 ( @xmath161 ) and @xmath162 ( @xmath163 ) .
the two - body problem has been solved using the numerov algorithm ( see ref .
@xcite for details ) and all the model parameters fixed in the description of the @xmath9 spectra @xcite . since the only relevant two - meson decay thresholds for these four - quark systems are those formed by two @xmath164 mesons , we summarize in table [ t1 ] the results obtained for the charmonium spectrum in ref .
@xcite compared with the experimental data quoted by the particle data group ( pdg ) @xcite .
it can be observed that the energy difference in the @xmath165wave @xmath166 multiplets due to the non - central terms is less than 50 mev . although these terms are known to play an important role in the description of the light @xmath9 states , their contribution becomes smaller as the mass of the heavy quark increases @xcite .
this allows to neglect them on the four - body sector , because they would only provide with a fine tune of the final results but making the solution of the four - body problem much more involved and time consuming . to analyze the stability of these systems against dissociation through strong decay , @xmath167 , parity ( @xmath168 ) , @xmath25parity ( @xmath35 ) , and total angular momentum ( @xmath32 )
must be preserved .
assuming that the decay takes place in a relative @xmath169wave for the final state , @xmath170 and @xmath171
. since the final states are also eigenstates of @xmath25parity one has @xmath172 .
once the meson masses have been obtained , the corresponding thresholds can be computed simply adding the mesons masses @xmath173 and @xmath174 , @xmath175 . in table [ t2 ]
we indicate the lowest two - meson decay threshold for each set of quantum numbers .
four - quark states will be stable under strong interaction if their total energy lies below all possible , and allowed , two - meson thresholds .
it is useful to define @xmath176 in such a way that if @xmath177 the four - quark system will fall apart into two mesons , while @xmath178 will indicate that such strong decay is forbidden and therefore the decay , if allowed , must be weak or electromagnetic , being its width much narrower .
let us first of all analyze the convergence of the expansion in terms of hyperspherical harmonics .
we show in fig .
[ f1 ] the variation of the energy of the @xmath18 and @xmath179 cases with @xmath156 , this behavior being similar for the other quantum numbers . from this figure
it can be seen that the convergence is slow , and the effective potential techniques @xcite are unable to improve it . in order to obtain a more adequate value for the energy we have extrapolated it according to the expression @xmath180 where @xmath181 , @xmath182 and @xmath183 are fitted parameters . in table [ t2b ]
we show the values obtained for @xmath181 for the two states shown in fig .
[ f1 ] as a function of the fitting range @xmath184 .
it can be checked that the values obtained for @xmath181 are stable within @xmath18510 mev for @xmath186 .
the values obtained for the @xmath183 parameter are very similar for all possible quantum numbers , being in all cases in the range @xmath187 , corresponding the lower limit to the positive parity states and the upper one to the negative parity states .
the origin of the difficulties to obtain a better convergence can be traced back to two different aspects of these systems . on the one hand ,
the hh formalism is better suited to deal with bound states , however , in this particular case we will show that most of the four - quark states are above the corresponding two - meson threshold . on the other hand ,
the color structure of a four - quark system is more involved than that of @xmath9 or @xmath188 systems @xcite , with two different possibilities to obtain a color singlet .
this makes possible the existence of repulsive contributions due to the color operator of the interacting potential , eqs .
( [ oge ] ) and ( [ confi ] ) .
although the total interaction is always confining , these deconfining terms make the convergence much slower .
we have studied all possible @xmath189 quantum numbers with @xmath23 .
we show in table [ t3 ] the results obtained for @xmath190 , the maximum value of @xmath156 calculated , the results obtained using the extrapolation of eq .
( [ extra ] ) , the corresponding threshold for each set of quantum numbers , and the value of @xmath191 , eq . ( [ delta ] ) .
a first glance to the results shows that there are two sets of quantum numbers , @xmath192 and @xmath193 , where the four - quark configuration is clearly below the corresponding two - meson threshold for dissociation .
the @xmath159 and @xmath194 states are very close to the threshold and all the other quantum numbers are far above the corresponding threshold . being all possible strong decays
forbidden they should be narrow states with typical widths of the order of a few mev .
it is also interesting to note that the quantum numbers @xmath1 and @xmath3 correspond to exotic states , those whose quantum numbers can not be obtained from a @xmath9 configuration , and therefore , if experimentally observed , they would be easily distinguished as a clear signal of pure non@xmath195 states .
the energies obtained for the negative parity states are , as expected , much higher than those of positive parity and therefore all of them are far above the corresponding thresholds , @xmath196 mev .
thus , they will fall apart immediately and therefore they should be very broad and difficult to detect .
let us also note that the experimental observation stating that a unit of angular momentum ( parity change ) in the @xmath188 and @xmath9 sector costs approximately 400@xmath197500 mev of excitation energy in all flavor sectors ( with some remarkable exceptions as the light scalar mesons @xcite ) has an equivalent in the four - quark sector , being in this case the energy difference due to the parity change @xmath198 mev .
the existence of non@xmath195 signals in the meson spectra has been the subject of an intensive debate @xcite .
to analyze whether the existence of bound states with exotic quantum numbers could be a characteristic feature of the heavy quark sector or it is also present in the light sector we show in fig .
[ f2 ] the value of @xmath191 as a function of the quark mass for all exotic quantum numbers and in fig .
[ f3 ] for the remaining positive parity states .
since each quantum number has a different threshold these figures should be interpreted carefully .
one should notice that the value @xmath199 in both figures corresponds to @xmath200 , and therefore , the fact that one state is below the others in these figures do not imply that its total mass would be smaller .
let us also note that since the heavy quarks are isoscalar states , the flavor wave function of the four heavy - quark states will be completely symmetric with total isospin equal to zero .
therefore , one should compare the results obtained in the light - quark case with a completely symmetric flavor wave function , i.e. , the isotensor states . in fig .
[ f3 ] we observe how one of the four - quark non - exotic positive parity states , the @xmath2 , becomes more bound when the quark mass is decreased , @xmath201 mev for @xmath202 mev .
the @xmath4 and @xmath19 states , that were slightly above the threshold for @xmath203 mev , increase their attraction when the quark mass is increased and only for masses above 3 gev , close to the bottom quark mass , may be bound .
with respect to the exotic quantum numbers , we observe in fig .
[ f2 ] that the @xmath204 and @xmath205 are not bound for any value of the quark mass .
the @xmath3 state decreases its binding when the quark mass diminishes and it becomes unbound for masses of the order of 500 mev , the strange quark mass . only the @xmath1 four - quark state becomes more deeply bound when the constituent quark mass decreases , and therefore only one bound state with exotic quantum numbers would remain in the light - quark sector .
taking the experimental mass for the threshold @xmath206 in the light - quark case , one can estimate the energy region where these states can be found , being @xmath207 mev , @xmath208 mev , @xmath209 mev , and @xmath210 mev .
note that the boson exchange potentials , that as explained in sect .
ii are present in the description of the light - quark sector , have not been considered in this analysis .
although these terms do play a role in the description of the four - quark state total energy , it was shown in ref .
@xcite that their suppression will not alter the relative order of the four - quark states .
the four - quark states with quantum numbers @xmath19 and @xmath2 have been analyzed with the complete constituent quark model including boson exchanges using a variational approach @xcite , obtaining 1004 mev and 1500 mev respectively , both in good agreement with our calculation and therefore giving confidence on our approach neglecting the boson exchange interactions in this particular analysis .
the states with exotic quantum numbers would be higher in energy , being @xmath211 mev , @xmath212 mev and @xmath213 mev .
there are experimental evidences for three states with exotic quantum numbers in the light - quark sector .
two of them are isovectors with quantum numbers @xmath214 named @xmath215 and @xmath216 , and one isotensor @xmath194 , the @xmath217 @xcite . based on the coupling of the @xmath215 to the @xmath218 decay channel a possible four - quark structure has been suggested @xcite while the strong @xmath219 coupling of the @xmath216 makes it a good candidate for a pure hybrid state @xcite .
the large mass obtained in our analysis for these quantum numbers makes doubtful the identification of any of them , and in particular of the @xmath215 , with a pure four - quark state , although a complete calculation with the proper flavor wave function is needed before drawing a definitive conclusion @xcite .
an alternative explanation of the @xmath215 in terms of low - energy rescattering effects has been proposed in ref .
@xcite while recent reanalysis of experimental data from e852 collaboration with improved statistics show no evidence of any exotic meson with a mass close to 1.6 gev @xcite .
concerning the @xmath217 , being its experimental mass 1600@xmath185100 mev , a possible tetraquark configuration seems likely .
in this work we have presented for the first time a generalization of the hh formalism to study systems made of quarks and antiquarks of the same flavor . for this purpose
we made use of standard techniques widely used in nuclear physics that have been adapted to include the color degree of freedom and to treat explicitly the @xmath25parity .
this formalism opens the door to an exact study of multiquark systems up to now described by means of different techniques , being the variational methods the most standard ones .
the particular color structure of four - quark systems makes the convergence slow but attainable .
the formalism has been applied to study the existence of @xmath23 four - charm quark states . for this purpose we made use of a well - established constituent quark model that properly accounts for the charmed meson and baryon spectra .
our results suggest the possible existence of three four - quark bound states with quantum numbers @xmath1 , @xmath3 and @xmath2 and masses of the order of 6515 , 6648 , and 6216 mev .
the two states with exotic quantum numbers , clearly below their corresponding two - meson threshold , should present narrow widths and , if produced , may be easily detected .
we have analyzed the variation of our results with the constituent quark mass . in the light - quark case
only the @xmath1 and @xmath2 quantum numbers remain bound , being the @xmath1 the lowest one with an energy close to 1800 mev .
the four - quark state with quantum numbers @xmath205 lies around 2900 mev , far from the experimental states @xmath215 and @xmath216 , therefore supporting a hybrid configuration or a more complicated structure in terms of non - resonant structures . a possible description of the @xmath217 as a
four - quark has also been justified .
the program we have started for and exact study of multiquark systems by means of the hh formalism will be accomplished by implementing the possibility of treating quarks of different masses .
when this is done we will have at our disposal a powerful method , imported from the nuclear physics , to study in an exact way systems made of any number of quarks and antiquarks coupled to a color singlet .
this work has been partially funded by ministerio de ciencia y tecnologa under contract no .
fpa2004 - 05616 , by junta de castilla y len under contract no .
sa-104/04 , and by generalitat valenciana under contract no .
gv05/276 .
t. barnes , f.e .
close , and h.j .
lipkin , phys .
d * 68 * , 054006 ( 2003 ) ; a.p .
szczepaniak , phys .
b * 567 * , 23 ( 2003 ) ; k. terasaki , phys .
d * 68 * , 011501(r ) ( 2003 ) ; h.y .
cheng and w.s .
hou , phys .
b * 566 * , 193 ( 2003 ) ; y .- q .
chen and x .- q .
li , phys .
. lett . * 93 * , 232001 ( 2003 ) ; t. browder , s. pakvasa , and a. petrov , phys .
b * 578 * , 365 ( 2004 ) ; j. vijande , f. fernndez , and a. valcarce , phys . rev .
d * 73 * , 034002 ( 2006 ) ; s.l .
olsen , int .
a * 20 * , 240 ( 2005 ) ; l. maiani , f. piccinini , a.d .
polosa , and v. riquer , phys .
d * 71 * , 014028 ( 2005 ) ; l. maiani , v. riquer , f. piccinini , and a.d .
polosa , phys .
d * 72 * , 031502 ( 2005 ) .
j. carlson , l. heller , and j.a .
tjon , phys . rev .
d * 37 * , 744 ( 1988 ) ; a.v .
manohar and m.b .
wise , nucl .
b * 399 * , 17 ( 1993 ) ; s. pepin , fl .
stancu , m. genovese , and j .-
richard , phys .
b * 393 * , 119 ( 1997 ) ; e. eichten _
et al . _ ,
d * 17 * , 3090 ( 1978 ) ; _ ibid _ d * 21 * , 313 ( e ) ( 1980 ) ; r.k .
bhaduri , l.e .
cohler , and y. nogami , nuovo cim .
a * 65 * , 376 ( 1981 ) ; n. isgur and g. karl , phys . today * 36 * , 36 ( 1983 ) . p.w .
stephenson , nucl .
b * 550 * , 427 ( 1999 ) ; f. knechtli and r. sommer , nucl .
b * 590 * , 309 ( 2000 ) ; o. philipsen and h. wittig , phys .
b * 451 * , 146 ( 1999 ) ; p.
de forcrand and o. philipsen , phys .
b * 475 * , 280 ( 2000 ) .
|
we present a generalization of the hyperspherical harmonic formalism to study systems made of quarks and antiquarks of the same flavor .
this generalization is based on the symmetrization of the @xmath0body wave function with respect to the symmetric group using the barnea and novoselsky algorithm .
the formalism is applied to study four - quark systems by means of a constituent quark model successful in the description of the two- and three - quark systems .
the results are compared to those obtained by means of variational approaches .
our analysis shows that four - quark systems with exotic @xmath1 and non - exotic @xmath2 quantum numbers may be bound independently of the mass of the quark . @xmath3 and
@xmath4 states become attractive only for larger mass of the quarks .
|
here , we give some further details on the three - site model given by eq .
( [ eqn : eq1 ] ) of the main text which in its explicit form reads @xmath95 the equations of motion are obtained straight forwardly by using heisenberg s equation of motion . the mechanical phases
@xmath33 evolve according to @xmath96 the optical amplitudes @xmath97 obey the following equations of motion @xmath98 and the optical phases @xmath99 @xmath100 as mentioned in the main text , for @xmath53 , i.e. , in the case of a static flux @xmath86 , and for @xmath60 a circulator behavior is expected , cf . ref .
@xcite . a solution to this system of coupled first order differential equations with initially one photon on site @xmath101 and a phase @xmath79
is shown in fig .
[ fig : circ ] a ) and b ) , where the circulator behavior is clearly visible , i.e. , the photon is moving counterclockwise around the triangular plaquette . .
the initial state has one photon on site @xmath101 . for @xmath102 ,
the initial excitation moves counterclockwise around the triangular plaquette shown in b ) . for @xmath91 , the initial excitation would move clockwise around the triangle .
c ) dispersion relation @xmath103 as a function of the phase @xmath86 .
] the hamiltonian of the three - site model can be diagonalized best after going into a rotating frame by applying the transformation @xmath104 to eq .
( [ eqn : s1 ] ) which leads to the hamiltonian @xmath105 since we are here interested in the eigenvalues , we can perform a gauge transform @xmath106 and make the following gauge choice @xmath107 which we can write as @xmath108 .
the hamiltonian can then be written as @xmath109 where we assumed periodic boundary conditions , i.e. , @xmath56 and for convenience we introduce @xmath110 and similarly for @xmath111 . by introducing normal modes @xmath112 and furthermore assuming equal tunneling amplitude @xmath113 and limit - cycle amplitude @xmath27
, the hamiltonian can be diagonalized @xmath114 where @xmath115 . as already pointed out in ref .
@xcite , for @xmath116 this setup shows the behaviour of a photon circulator .
since the photons eventually decay , we add photon loss and drive to the system .
this is best done starting with the hamiltonian ( [ eqn : s1 ] ) and adding a driving term @xmath117 . going into a frame rotating with the driving frequency @xmath118 , we obtain @xmath119 including dissipation of the photons at a rate @xmath18 and neglecting effects due to quantum noise , the equations of motion for the photons become @xmath120 with @xmath121 the equations of motion for the phases @xmath33 are unchanged . in fig .
[ fig : okdiss ] we show the optical amplitude @xmath122 as a function of the flux @xmath86 and the drive frequency @xmath118 .
figure [ fig : okdiss ] almost resembles the eigenfrequencies @xmath103 in fig .
[ fig : circ ] c ) which we obtained from diagonalizing the hamiltonian . .
middle : @xmath123 .
right : @xmath124 .
here we only drive site @xmath63 , i.e. , @xmath125 and choose @xmath126 , and @xmath127 . ]
on a two - dimensional triangular lattice we denote dynamical variables on lattice site @xmath128 with a subscripts @xmath129 and dynamical variables on the directed links from site @xmath128 to site @xmath130 by a subscript @xmath131 .
the equations of motion for the mechanical phases @xmath132 on the lattice links are @xmath133 the optical amplitudes @xmath134 obey the equations of motion @xmath135 and the optical phases @xmath136 on a lattice site evolve according to @xmath137 a driving term on a particular site and dissipation of the photons on each site can be added straightforwardly , as also done for the three - site model .
as mentioned in the main text , the links are directed : a photon tunneling from a site with a low ( high ) on - site frequency to a site with a high ( low ) on - site frequency absorbs ( emits ) a phonon . these processes have to be resonant which could be achieved by distributing the photon and phonon frequencies in the right way .
for instance , the frequencies can be arranged as shown in fig . [
fig : fe ] .
|
artificial gauge fields for neutral particles such as photons , recently attracted a lot of attention in various fields ranging from photonic crystals to ultracold atoms in optical lattices to optomechanical arrays . here
we point out that , among all implementations of gauge fields , the optomechanical setting allows for the most natural extension where the gauge field becomes dynamical .
the mechanical oscillation phases determine the effective artificial magnetic field for the photons , and once these phases are allowed to evolve , they respond to the flow of photons in the structure .
we discuss a simple three - site model where we identify four different regimes of the gauge - field dynamics .
furthermore , we extend the discussion to a two - dimensional lattice .
our proposed scheme could for instance be implemented using optomechanical crystals . *
_ introduction. _ * gauge theories are very elementary concepts in physics . today we know that the fundamental interactions between particles are mediated through gauge fields .
the most common known gauge theory is classical electrodynamics which is described by maxwell s equations . in quantum mechanics the simplest but non trivial setting is a charged quantum particle moving in a classical externally prescribed electromagnetic background field giving rise to phenomena like the aharonov - bohm effect .
field theories such as quantum electrodynamics ( qed ) and quantum chromodynamics ( qcd ) are gauge theories where the gauge fields are dynamical in their own right and which interact with the matter fields , thereby mediating interactions among the matter particles .
recently , intriguing proposals showing how to synthesize dynamical gauge fields have been put forward .
these proposals include the generation of abelian as well as non - abelian gauge fields realized with ultracold atoms in optical lattices @xcite or with superconducting circuits @xcite .
these settings can be viewed as a powerful tool for the quantum simulation of dynamical gauge theories such as qed and qcd on a lattice @xcite . in this article
we study the most basic phonon - assisted photon tunneling process which is due to the optomechanical interaction .
we are interested in the coupled photon - phonon dynamics and we show that in the optomechanical setting , phonons can act as a _ dynamical _ gauge field for photons .
as we will show , in the optomechanical setting , these dynamical gauge fields arise in quite a natural manner . in general ,
optomechanics describes the interaction of light and mechanical motion @xcite .
the prototypical optomechanical setting consists in a fabry - prot cavity where one of the mirrors is free to oscillate . due to the radiation pressure force the light inside the cavity interacts with the mirror s motion .
tremendous experimental progress has been made during the last years , with achievements such as cooling a nanomechanical oscillator to its motional ground state @xcite and position measurements below the standard quantum limit @xcite , to name only a few examples . mechanically and/or optically coupling several optomechanical systems
leads to interesting new physics .
for instance , setups consisting of only a few optical and mechanical modes allow for nonreciprocal devices for photons @xcite .
furthermore , larger optomechanical arrays , i.e. , one- or two - dimensional arrangements of coupled optomechanical systems are promising candidate systems for studying many - body physics of photons or phonons @xcite .
recently , it has been shown that optomechanical arrays also allow for the generation of artificial gauge fields for photons @xcite and phonons @xcite .
the optomechanical implementation complements other proposals for generating artificial gauge fields for photons @xcite and ultracold atoms in optical lattices @xcite . as mentioned , _
static _ gauge fields for photons can also be realized using a variety of other methods , even though they are not necessarily as tunable as the optomechanical approach . in this article , however , we want to exploit a unique aspect of the optomechanical case that is not present in any of the other approaches that are based on geometry or externally applied periodic modulation .
this is the fact that the mechanical vibrations ( which generate the artificial gauge field ) have oscillation phases that can evolve in response to the light field dynamics .
thus , our aim is to promote the static gauge fields to become dynamical gauge fields and investigate back - action effects of the dynamical gauge field on the hopping photons in the intriguing driven and dissipative setting of an optomechanical array .
we introduce a simple model which gives rise to acoustical dynamical gauge fields for optical `` matter '' fields . to be more precise , we consider a photonic lattice ( representing the matter fields ) and artificial gauge fields ( phonons ) which can be attributed to directed links between two sites of the photonic lattice .
such a system could for instance be implemented in optomechanical crystal structures @xcite or disk resonator arrays @xcite .
( light blue ) coupling to phonons @xmath0 ( orange ) . here
, the optical frequencies increase around the triangle , @xmath1 . as a consequence
the links are directed , i.e. , photons hopping in direction of the orange arrow absorb a phonon .
b ) possible realization making use of the so - called `` modulated link '' scheme ( see main text ) .
dark blue ( green ) dots depict the optical ( mechanical ) modes of the intermediate sites .
intermediate optical and mechanical modes couple optomechanically to form an optomechanical cell ( dashed box ) .
c ) two - dimensional triangular lattice .
the arrows indicate the link direction . ]
* _ model. _ * the crucial ingredient in our model is the phonon - assisted photon tunneling that can be generated by the optomechanical interaction : a photon @xmath2 in a lattice hopping from site to site is accompanied by the coherent emission or absorption of a phonon @xmath0 . for a photon tunneling between sites @xmath3 and @xmath4 , this is described by @xmath5 ( and the opposite process ) .
such a setting can be described by the hamiltonian @xmath6 we set @xmath7 . here
, @xmath4 denotes a lattice site , and @xmath8 is the index for a _ directed _ link from @xmath9 to @xmath10 .
when a photon hops in the direction of the link , it absorbs a phonon .
photons ( phonons ) have frequencies @xmath11 ( @xmath12 ) , and @xmath13 are the coupling constants for the phonon - assisted photon tunneling process . for later use , we introduce the notation @xmath14 and note that we use @xmath15 and @xmath16 interchangeably . in eq .
( [ eqn : eq1 ] ) we made use of the rotating wave approximation which is valid for @xmath17 , where @xmath18 is the photon decay rate .
as pointed out previously @xcite , non - reciprocity in photon transport can be engineered by introducing coherent inelastic transitions induced by mechanical vibrations . however , in contrast to ref . @xcite we will treat the vibrations as dynamical degrees of freedom , as explained below .
regardless of the details of the scheme , in order to make the inelastic processes resonant we need to ensure that the nearest neighbor on - site photon frequencies @xmath11 differ by the corresponding link phonon frequency @xmath19 , i.e. , @xmath20 , with the link direction from site @xmath3 to @xmath4 .
this leads to _ directed _ links .
generally , a photon tunneling from a site with a low ( high ) on - site frequency to a site with a high ( low ) on - site frequency absorbs ( emits ) a phonon .
since the photons will eventually decay , we include the effects of dissipation and driving of the photons by adding a driving term @xmath21 to the hamiltonian ( [ eqn : eq1 ] ) and by including dissipation of the photons at a rate @xmath18 ( for details we refer to the supplemental material @xcite ) .
this effective model can be realized in an optomechanical setting by building on the `` modulated link '' scheme that had been proposed in ref .
@xcite to generate _
static _ artificial magnetic fields .
we briefly review this scheme for one of the links of the optomechanical setup depicted in fig .
[ fig : om_schematics ] b ) .
we explain why such a modulated link reduces to the link of the effective model of eq .
( [ eqn : eq1 ] ) , cf . fig .
[ fig : om_schematics ] a ) .
the link between two optical modes @xmath3 and @xmath4 is realized with an intermediate optical mode @xmath22 which couples optomechanically to a mechanical mode @xmath23 .
consequently , the link forms a single optomechanical cell . in the original `` modulated link ''
scheme @xcite the mechanical mode is externally driven into a large amplitude state @xmath24 which leads to a modulation of the frequency of the intermediate optical mode @xmath25 , where @xmath26 is the single - photon optomechanical coupling strength . however , in contrast to @xcite , we will here assume the mechanical oscillator to be undergoing self - sustained optomechanical oscillations @xcite instead of being externally driven .
thus it behaves as a limit - cycle oscillator with a fixed amplitude @xmath27 and a free phase @xmath28 . in any case , in order to enable a photon to tunnel along such a link , the mechanical frequency has to match the difference of optical frequencies , as explained above .
we note that other microscopic implementations are possible .
for example , one might have a mechanical mode that directly couples to the hopping between optical modes , as has been worked out in detail for optomechanical crystals @xcite .
this would be connected to the `` wavelength conversion scheme '' discussed in ref .
@xcite .
* _ gauge field dynamcis. _ * the corresponding equations of motion for the effective hamiltonian ( [ eqn : eq1 ] ) for the operators @xmath29 and @xmath30 can be obtained using heisenberg s equations of motion . here
we will focus on the classical dynamics of the model , i.e. , the limit of large coherent photon and phonon amplitudes .
this is the most relevant regime for most of the current optomechanical setups ( due to the small single - photon coupling strength @xmath26 ) @xcite .
thus , we decompose the expectation values of the photon and phonon operators into a classical amplitude and a phase , @xmath31 and @xmath32 . from the full quantum heisenberg equations of motion ,
the equations for the mechanical phases @xmath33 become @xmath34 where the link direction is from @xmath3 to @xmath4 , while the optical amplitudes @xmath35 obey the following equations of motion @xmath36 and the optical phases @xmath37 evolve according to @xmath38 here and in the following we assume that the amplitudes of the limit - cycle oscillations @xmath39 are a constant of motion , i.e. , @xmath40 .
this regime can be reached by working with self - induced optomechanical oscillators ( pumped by a blue - detuned laser ) sufficiently above threshold @xcite . furthermore , we introduced @xmath41 and used @xmath42 .
an important point we want to make at this stage is the invariance of the equations of motion under the following _ local _
@xmath43 gauge transformation @xmath44 meaning that the observed evolution of the light _
intensity _ will not change , independent of the choice of @xmath45 .
we assumed a static gauge choice @xmath46 ; otherwise eqs .
( [ eqn : g1 ] ) and ( [ eqn : g2 ] ) would have to be supplemented by a change in frequencies : @xmath47 and @xmath48 .
we now ask under which conditions the gauge field will display nontrivial dynamics .
it turns out that the time occurring in eqs .
( [ eqn : eqm1 ] ) , ( [ eqn : eqm2 ] ) , and ( [ eqn : eqm3 ] ) can be rescaled by @xmath49 . after doing this
, we observe that the entire dynamics only depends on the dimensionless ratio of @xmath50 ( where @xmath51 is proportional to the laser drive amplitude ) .
for the limit @xmath52 , we expect that the oscillation phases ( i.e. , the magnetic field ) will not be affected by the hopping photons , rather only defining a static magnetic field pattern .
this can be seen from eq .
( [ eqn : eqm1 ] ) : for @xmath53 the second term , which provides the coupling to the hopping photons , can be neglected . however ,
if @xmath50 is large , we expect back - action of the hopping photons on the phonons leading to intriguing coupled dynamics of the gauge field .
* _ three - site model. _ * first , we will study the case of three sites .
the resulting effective model for photons @xmath29 on sites @xmath54 and phonons @xmath15 on links @xmath55 is depicted in fig .
[ fig : om_schematics ] a ) .
note that for three sites we define @xmath56 .
for definiteness , we will assume @xmath57 .
for the present three - site model , the gauge freedom implies that only the gauge invariant flux @xmath58 i.e. , the sum of phases around the triangular plaquette , affects the dynamics of the photons .
we want to mention that in the regime where the phonons are not influenced by the photons ( @xmath53 and consequently @xmath59 ) the hamiltonian for the three - site model can be diagonalized and the setup can act ( for @xmath60 ) as a photon circulator @xcite . in the following we turn to the intriguing case where the photons interact with the phonons which produces a dynamical gauge field . here
, we discuss the driven and dissipative setting and furthermore choose equal tunneling amplitudes @xmath61 and mechanical amplitudes @xmath62 . in the following we consider an example with a resonant drive on site @xmath63 only , i.e. , @xmath64 , @xmath65 and @xmath66 .
as pointed out above , without drive and dissipation the dynamics only depends on the ratio @xmath50 . as a consequence ,
the four parameters involved here , for the driven case @xmath67 , can be combined into just two dimensionless parameters , @xmath68 and @xmath69 .
the resulting `` phase diagram '' for the flux dynamics as a function of these two parameters is displayed in fig .
[ fig : fig2 ] a ) , and has been obtained from direct numerical simulations .
these simulations reveal four distinct regimes . in regimes
i and ii the flux @xmath70 ( after some transient behavior ) approaches a stationary value of either @xmath71 or different from it , respectively . in regimes
iii and iv , the flux @xmath70 is not stationary but shows a dynamical behavior even in the long - time limit of the driven and dissipative case .
most interestingly , the flux dynamics @xmath70 in this regime can either show periodic or chaotic behavior , region iii and iv in fig .
[ fig : fig2 ] a ) , respectively .
figures [ fig : fig2 ] b ) and c ) show cuts along the red dashed lines marked in the phase diagram .
the cuts indicate a continuous phase transition from phase i to ii . in fig .
[ fig : fig2 ] d ) and e ) , we show two examples of the phase space @xmath72 in regimes iii and iv . already at this level we can distinguish periodic [ fig .
[ fig : fig2 ] d ) ] from chaotic [ fig . [
fig : fig2 ] e ) ] dynamics .
a more involved characterization can be done with the help of a bifurcation diagram . to this end
, we show the value of the flux @xmath70 evaluated at the zero crossings of @xmath73 , in the long - time limit , as a function the mechanical amplitude @xmath27 in fig .
[ fig : fig2 ] f ) .
this bifurcation diagram allows us to distinguish the periodic [ region iii in fig .
[ fig : fig2 ] a ) ] from the chaotic [ region iv in fig . [
fig : fig2 ] a ) ] flux dynamics within the whole phase diagram for @xmath70 .
analytical progress can be made in the limit where the photon dynamics is much faster than the phonon dynamics .
we are then able to apply a born - oppenheimer approximation and adiabatically eliminate the photons .
to be more precise , we solve @xmath74 where @xmath75 a_{j } - i e_{j } - i j b \sum_{k \neq j } e^{-i\phi_{jk } } a_{k } \ , , { \nonumber}\end{aligned}\ ] ] and use this instantaneous solution to eliminate @xmath76 from the equations of motion for the mechanical phases @xmath33 ( for details we refer to the supplemental material @xcite ) . in the case of a resonant ( @xmath66 ) drive on site @xmath63 , this approximation leads to the following equation of motion for the flux : @xmath77^{2 } + [ 16 \cos(\phi)]^{2 } } \label{eqn : dp } \ , .\end{aligned}\ ] ] from eq .
( [ eqn : dp ] ) we find @xmath71 which shows very good agreement with the exact numerical long - time dynamics in regime i. this approach fails to reproduce the flux dynamics @xmath70 in the other two regimes since there we are not able to adiabatically eliminate the photons .
we also want to mention that both @xmath78 and @xmath79 are fixed points of eq .
( [ eqn : dp ] ) .
it turns out that for a resonant drive on site @xmath63 , @xmath79 is an unstable fixed point of eq .
( [ eqn : dp ] ) .
the asymmetry between @xmath80 and @xmath81 is due to the breaking of translational invariance , necessarily produced by the link directions .
in contrast , a resonant drive on site @xmath82 would have @xmath83 as a stable fixed point .
* _ lattices. _ * we extend the three - site model to two - dimensional lattices and illustrate the dynamical behaviour on a triangular lattice [ fig .
[ fig : om_schematics ] c ) ] .
we focus on a driven and dissipative situation .
going from three sites to a lattice , a significant new feature comes into play : the artificial dynamical magnetic field produced by the phonons can now exert a lorentz force that bends the path of any light - beam propagating in the two - dimensional array .
therefore , we end up with a dynamical interplay where the flow of the photons changes the spatial distribution of the magnetic flux density which then acts back on the dynamics of the light field . in the following we assume equal mechanical amplitudes @xmath84 and tunneling amplitudes @xmath85 .
the frequencies of the optical modes determine the link directions .
we choose a scenario with the link directions as depicted in fig .
[ fig : om_schematics ] c ) , cf . supplemental material @xcite for details .
we note that this is not the only possible choice .
in fact the intricate photon and phonon dynamics depends on the link pattern .
we illustrate the nonlinear structure formation in this model for the case of having only a single site illuminated by a laser . on each site ( top row ) and
flux @xmath86 per plaquette ( bottom row ) at different times .
the simulation was carried out on a @xmath87 triangular lattice , whose left half is shown here .
the resonant drive is on the site on the corner and the initial flux per plaquette is @xmath88 .
no steady state of either photons or phonons is reached .
we chose @xmath89 , @xmath90 . ] in fig .
[ fig : drivedissdyn ] we show the temporal evolution of the light intensity ( top row ) as well as the magnetic field ( bottom row ) on the lattice . at first the photons experience a static magnetic field which is set by the initial phases of the mechanical oscillations ( here chosen such that the initial flux is @xmath91 ) and
start to move along the edge . due to the back - action of the photons ( which primarily move along the edge ) on the phonons , the flux per plaquette changes .
this in turn leads to a reconfiguration of the magnetic field .
even though the photons live only for a short time @xmath92 before escaping the structure , the system develops a spatial `` memory '' in the form of the mechanical oscillation phases , where previous photons leave their imprint .
the photons then react to this new magnetic field configuration . in this specific example
, the photons reverse their direction of motion . in the simulations performed here ,
the system does not reach a steady state in the long - time limit .
collective behavior such as synchronization and pattern formation of mechanical limit - cycle oscillators have recently been theoretically studied in optomechanical arrays @xcite . in connection to these investigations , further studies on gauge field dynamics in optomechanics
could for instance address questions on synchronization and possible dynamical pattern formation of the magnetic field .
* _ conclusion. _ * we have shown that dynamical gauge fields in optomechanical arrays arise quite naturally . the evolving mechanical phases , which respond to the flow of the photons , represent a dynamical gauge field for the latter .
already the three - site model shows intriguing dynamics which leads to a rather complex phase diagram for the flux dynamics . with experiments pushing towards multi - mode optomechanical setups ,
the three - site model seems feasible to be realized in the near future and would pave the way for further studies of dynamical gauge fields in optomechanical arrays . _
acknowledgements : _ we acknowledge helpful discussions with roland lauter , and thank vittorio peano and talitha weiss for a careful reading of the manuscript .
this work was financially supported by the marie curie itn cqom and the erc optomech
. 99 k. osterloh , m. baig , l. santos , p. zoller , and m. lewenstein , phys .
rev .
lett . * 95 * , 010403 ( 2005 ) .
j. ruseckas , g. juzelinas , p. hberg , and m. fleischhauer , phys .
rev .
lett . * 95 * , 010404 ( 2005 ) .
p. hauke , o. tieleman , a. celi , c. lschlger , j. simonet , j. struck , m. weinberg , p. windpassinger , k. sengstock , m. lewenstein , and a. eckardt , phys .
rev .
lett . * 109 * , 145301 ( 2012 ) .
d. banerjee , m. dalmonte , m. mller , e. rico , p. stebler , u. j. wiese , and p. zoller , phys .
rev .
lett . * 109 * , 175302 ( 2012 ) .
e. zohar , j. i. cirac , and b. reznik , phys .
rev .
a * 88 * , 023617 ( 2013 ) .
m. j. edmonds , m. valiente , g. juzelinas , l. santos , and p. hberg , phys .
rev .
lett . * 110 * , 085301 ( 2013 ) .
d. marcos , p. rabl , e. rico , and p. zoller , phys .
rev .
lett . * 111 * , 110504 ( 2013 ) .
d. marcos , p. widmer , e. rico , m. hafezi , p. rabl , u. j. wiese , and p. zoller , annals of physics * 351 * , 634 ( 2014 ) .
u. j. wiese , ann .
der phys .
* 525 * , 777 ( 2013 ) .
e. zohar , j. i. cirac , and b. reznik , arxiv:1503.02312 ( 2015 ) .
m. aspelmeyer , t. j. kippenberg , and f. marquardt , rev .
mod .
phys . *
86 * , 1391 ( 2014 ) .
j. d. teufel , t. donner , d. li , j. w. harlow , m. s. allman , k. cicak , a. j. sirois , j. d. whittaker , k. w. lehnert , and r. w. simmonds , nature * 475 * , 359 ( 2011 ) .
j. chan , t. p. mayer alegre , a. h. safavi - naeini , j. t. hill , a. krause , s. grblacher , m. asperlmeyer , and o. painter , nature * 478 * , 89 ( 2011 ) .
j. d. teufel , t. donner , m. a. castellanos - beltran , j. w. harlow , and k. w. lehnert , nat .
nanotechnol . * 4 * , 820 ( 2009 ) .
s. manipatruni , j. t. robinson , and m. lipson , phys .
rev .
lett . * 102 * , 213903 ( 2009 ) .
s. j. m. habraken , k. stannigel , m. d. lukin , p. zoller , and p. rabl , new j. phys .
* 14 * , 115004 ( 2012 ) .
m. hafezi and p. rabl , opt .
express * 20 * , 7672 ( 2012 ) .
m. bhattacharya and p. meystre , phys .
rev .
a * 78 * , 041801 ( 2008 ) .
d. e. chang , a. h. safavi - naeini , m. hafezi , and o. painter , new j. phys .
* 13 * , 023003 ( 2011 ) .
a. xuereb , c. genes , and a. dantan , phys .
rev .
lett .
* 109 * 223601 ( 2012 ) .
a. tomadin , s. diehl , m. d. lukin , p. rabl , and p. zoller , phys .
rev .
a * 86 * , 033821 ( 2012 ) .
m. schmidt , m. ludwig , and f. marquardt , new j. phys .
* 14 * , 125005 ( 2012 ) .
u. akram , w. munro , k. nemoto , and g. j. milburn , phys .
rev .
a * 86 * , 042306 ( 2012 ) .
w. chen and a. a. clerk , phys .
rev . a * 89 * , 033854 ( 2014 ) .
m. schmidt , v. peano , and f. marquardt , new j. phys .
* 17 * , 023025 ( 2015 ) .
m. schmidt , s. keler , v. peano , o. painter , and f. marquardt , optica * 2 * , 635 ( 2015 ) .
v. peano , c. brendel , m. schmidt , and f. marquardt , phys .
rev .
x * 5 * , 031011 ( 2015 ) .
s. raghu and f. d. m. haldane , phys .
rev .
lett . * 100 * , 013904 ( 2008 ) .
z. wang , y. chong , j. d. joannopoulos , and m. soljai , nature * 461 * , 772 ( 2009 ) .
j. koch , a. a. houck , k. le hur , and s. m. girvin , phys .
rev .
a * 82 * , 043811 ( 2010 )
. r. o. umucallar and i. carusotto , phys .
rev .
a * 84 * , 043804 ( 2011 ) .
m. hafezi , e. a. demler , m. d. lukin , and j. m. taylor , nat .
phys .
* 7 * , 907 ( 2011 ) .
k. fang , z. yu , and s. fan , nature photon .
* 6 * , 782 ( 2012 ) .
l. lu , j. d. joannopoulos , and m. soljai , nature photon .
* 8 * , 821 ( 2014 ) .
m. hafezi , s. mittal , j. fan , a. migdall , and j. m. taylor , nature photon .
* 7 * , 1001 ( 2013 ) .
s. mittal , j. fan , s. faez , a. migdall , j. m. taylor , and m. hafezi , phys .
rev .
lett . * 113 * , 087403 ( 2014 ) .
d. jaksch and p. zoller , new j. phys .
* 5 * , 56 ( 2003 ) .
a. s. srensen , e. demler , and m. d. lukin , phys .
rev .
lett . * 94 * , 086803 ( 2005 ) .
y.j .
lin , r. l. compton , k. jimnez - garca , j. v. porto , and i. b. spielman , nature * 462 * , 628 ( 2009 ) .
m. aidelsburger , m. atala , s. nascimbne , s. trotzky , y.a .
chen , and i. bloch , phys .
rev .
lett . * 107 * , 255301 ( 2011 ) .
m. aidelsburger , m. atala , m. lohse , j. t. barreiro , b. paredes , and i. bloch , phys .
rev .
lett . * 111 * , 185301 ( 2013 ) .
g. jotzu , m. messer , r. desbuquois , m. lebrat , t. uehlinger , d. greif , and t. esslinger , nature * 515 * , 237 ( 2014 ) .
m. eichenfield , j. chan , r. m. camacho , k. j. vahala , and o. painter , nature * 462 * , 78 ( 2009 ) .
a. h. safavi - naeini and o. painter , opt .
express * 18 * , 14926 ( 2010 ) .
a. h. safavi - naeini , t. p. mayer alegre , m. winger , and o. painter , appl .
phys .
lett . * 97 * , 181106 ( 2010 ) .
e. gavartin , r. braive , i. sagnes , o. arcizet , a. beveratos , t. j. kippenberg , and i. robert - philip , phys .
rev .
lett . * 106 * , 203902 ( 2011 ) .
j. chan , t. p. mayer alegre , a. h. safavi - naeini , j. t. hill , a. krause , s. grblacher , m. aspelmeyer , and o. painter , nature * 478 * , 89 ( 2011 ) .
a. h. safavi - naeini , j. t. hill , s. meenehan , j. chan , s. grblacher , and o. painter , phys .
rev .
lett . * 112 * , 153603 ( 2014 ) .
m. zhang , s. shah , j. cardenas , and m. lipson , arxiv:1505.02009 ( 2015 ) .
see supplemental material at [ link ] for details .
a. h. safavi - naeini and o. painter , new j. phys .
* 13 * , 013017 ( 2011 ) .
f. marquardt , j. g. e. harris , and s. m. girvin , phys .
rev .
lett . * 96 * , 103901 ( 2006 ) .
g. heinrich , m. ludwig , j. qian , b. kubala , and f. marquardt , phys .
rev .
lett . * 107 * , 043603 ( 2011 ) .
m. ludwig and f. marquardt , phys .
rev .
lett . * 111 * , 073603 ( 2013 ) .
r. lauter , c. brendel , s. j. m. habraken , and f. marquardt , phys .
rev .
e * 92 * , 012902 ( 2015 ) .
stefan walter @xmath93 and florian marquardt @xmath94 + _ _ + _ _
|
1 . solar neutrino spectrum in the ssm [ 1 ] .
2 . typical oscillation factor .
3 . typical lwvo solutions [ 8 ] .
eigenvalues of propagation matrix versus density @xmath37 .
5 . the msw bathtub ; suppression factor @xmath95 versus @xmath96 . 6 .
msw solutions [ 16 ] .
typical exotic neutral current effects on msw solutions [ 19 ] .
modulation factors for @xmath7 neutrino spectrum .
electron spectra due to @xmath7 neutrinos [ 16 ] .
|
this is an invited review talk , presented at the international conference on non - accelerator particle physics ( icnapp-94 ) , bangalore , india , 2 - 9 january 1994 .
6.5 in 8.5 in .25 in # 1#20=1=to0@xmath00=1=to0#2#1 - 01
@=11 tempcntc citex[#1]#2@fileswauxout tempcnta@tempcntb@neciteaciteforciteb:=#2citeo#1 citeotempcnta > tempcntbciteacitea , tempcnta = tempcntbtempcnta
@=12 = cmssbx10 scaled 2 to * solar neutrino oscillations * + r.j.n .
phillips + _ rutherford appleton laboratory , chilton , didcot , oxon ox11 0qx , england _
* 1 .
introduction*. standard solar models ( ssm ) predict the @xmath1 flux of fig.1 with some uncertainties@xcite . measurements by capture in @xmath2 @xcite , @xmath3 scattering@xcite and capture in @xmath4 @xcite , with differing @xmath5 thresholds , find three different deficits : @xmath6 & observation / ssm[2 ] \\
\nu -e & 7.5 mev & 0.51\pm .07\pm .07 & 0.66\pm .09\pm .16 \\ ^{37}cl & 0.81 mev & 0.29\pm .03\pm .04 & 0.36\pm .04\pm .08 \\
^{71}ga & 0.24 mev & 0.62\pm .10\pm .03 & 0.67\pm .11\pm .04 \\ \end{array}\ ] ] where the first error is experimental , the second is from the ssm .
these numbers suggest a differential suppression , with the top and bottom of the accessible range less suppressed than the middle .
they pose the solar neutrino problem 1994 .
re - tuning the solar model gives no easy solution@xcite .
a lower central temperature would suppress @xmath7 production and the @xmath3 rate , but to explain @xmath2 rates the @xmath8 line must then be obliterated - a bit unlikely given that @xmath7 is made from @xmath8 .
neutrino oscillations offer several possible explanations , that i briefly compare . *
2 . long wavelength vacuum oscillations ( lwvo)*. suppose the weak eigenstate @xmath1 , emitted by @xmath9-decays in the sun , is actually a superposition of two mass eigenstates : @xmath10 with @xmath11 .
the mass eigenstates propagate independently with time t , each picking up a different phase factor @xmath12 , so that after a distance @xmath13 the projection back onto @xmath1 becomes @xmath14 $ ] .
the probability that this evolved state can interact like @xmath1 is then @xmath15 where @xmath16 .
figure 2 illustrates this oscillatory probability .
for @xmath17 there is negligible effect ; for values @xmath18 there are resolvable oscillations ; for values @xmath19 the oscillations are averaged in practice , either by source / detector size or by energy resolution . averaged 2-neutrino oscillations
suppress by at most 1/2 , but n - neutrino mixing can give 1/n ; however this suppression is flat and therefore unsuited to the 1994 solar problem . on the other hand , resolved oscillations provide a strongly varying suppression .
try overlaying figs.1 and 2 ( they have the same horizontal log scale ) .
if we tune @xmath20 such that the first minimum falls around a few mev , the kamiokande @xmath21 rate will be somewhat suppressed ; if we fine - tune to put the @xmath22 line in a minimum , the @xmath2 rate will be somewhat more suppressed ; meanwhile the @xmath4 rate suffers less , since its dominant pp neutrinos encounter only average suppression .
we clearly have the makings of one or more solutions here , with these just - so " oscillations@xcite .
figure 3 shows typical recent fits @xcite in the @xmath23 ) parameter plane ; the disconnected regions put the @xmath8 line in different minima of p. note that for @xmath24 mixing , @xmath25 scattering contributes a bit to the kamiokande signal and helps to explain why it is less suppressed than @xmath2 ; @xmath26 sterile flavour mixing lacks this help and is harder to fit .
two special features arise from lwvo resolved oscillation patterns @xcite .
+ i ) there is an oscillatory modulation on the shape of the high- energy @xmath7 spectrum contribution .
the shape ( though not the magnitude ) of this ssm component is model - independent ; the modulation would be detectable at sno@xcite and super - kamiokande .
see figs .
8,9 at the end .
+ ii ) the @xmath8 line with fixed @xmath5 has oscillatory strength , because the earth - sun distance @xmath27 has small seasonal variations .
this line strength could be measured directly by borexino@xcite ; the effect is diluted in @xmath2 and @xmath4 signals . + ( present @xmath28 , @xmath29 and @xmath30 data constrain ( i ) and ( ii ) rather weakly ) .
but mixing @xmath1 with @xmath31 or @xmath32 affects @xmath1 spectra from supernovae ; sn1987a data may disfavour large @xmath33 @xcite , including solutions like fig.3a . *
3 . oscillations in matter*. coherent forward scattering in matter generates a refractive index and affects propagation@xcite .
z - exchange processes are the same for @xmath1 , @xmath31 , @xmath32 , generating a common phase that can be ignored , but w - exchange contributes only to @xmath34 scattering and significantly changes the propagation equation : @xmath35 where @xmath36 is the vacuum mixing angle , @xmath37 is the electron number density and @xmath38 or @xmath32 ( for sterile @xmath39 see later ) . diagonalizing the propagation matrix above
, we find that the mixing angle in matter @xmath40 depends on @xmath41 : @xmath42 if @xmath43 the mixing is enhanced ; it becomes maximal ( @xmath44 ) where the denominator vanishes - sometimes called a resonance .
as neutrinos travel out from the solar core , the mixing angle @xmath40 and the matter - propagation eigenstates @xmath45 change continuously .
this gives a possibility for efficient @xmath46 conversion via adiabatic level crossing ( the msw effect@xcite ) .
suppose that @xmath41 is far above the resonance value at the point of @xmath1 creation in the solar core ( i.e. @xmath47 ) ; then @xmath48 here .
if subsequent propagation is adiabatic , the local eigenstate components are essentially preserved : @xmath49 .
emerging from the sun , the dominant @xmath50 component becomes the vacuum mass eigenstate @xmath51 if the vacuum mixing angle @xmath36 is small ; thus initial @xmath1 ends up as mostly @xmath31 .
fig.4 shows how the two eigenvalues @xmath52 of the propagation matrix behave versus @xmath37 ; solid lines give the case of no mixing , @xmath53 ; dashed lines show how the eigenvalues cross over when mixing is present .
if @xmath37 changes slowly enough for the mixing to act ( adiabatically ) , the physical state follows the full eigenstates ( dashed lines ) ; but if @xmath54 changes too suddenly , the physical state follows the unmixed eigenstates ( solid lines ) .
there are 2 conditions for adiabatic level crossing .
+ ( i ) central density is above the resonance value : @xmath55 \simeq 10 ^ 5,\ ] ] ( ii ) density changes slowly enough near the resonance for adiabaticity : from the landau - zener approximation we obtain@xcite @xmath56cos2\theta \simeq 3 \times 10 ^ 8 sin^22\theta.\ ] ] the detailed consequences require big calculations , but we need no computer to see the main features .
the msw effect gives a bathtub - shaped suppression factor ; see fig.5 .
the steep left - hand end is determined by the resonance - crossing condition ( i ) ; the sloping right - hand end covers the range where adiabaticity breaks down , determined by condition ( ii ) .
we can choose almost any bathtub we please , versus energy @xmath5 , by selecting @xmath57 to get the left - hand end and @xmath58 to get the length ( and also the depth ) of the bathtub .
however , condition ( ii ) excludes msw effects in the lwvo region .
notice also that efficient @xmath59 conversion does not require big vacuum mixing @xmath36 ; on the contrary , the best conversion is with small @xmath36 .
the msw bathtub offers an immediate explanation of the apparent differential suppression of the @xmath1 spectrum : let the sloping end lie across the @xmath3 scattering range @xmath60 ( moderate @xmath3 suppression ) ; let the flat bottom lie across the rest of the @xmath2 capture range @xmath61 ( more @xmath2 suppression ) ; let the steep end fall near @xmath62 at the top of the @xmath63 spectrum contribution ( less @xmath4 suppression ) .
this simple prescription leads to the best msw solution : @xmath64 with @xmath65 , fig.6a shows a typical recent fit@xcite .
there is also a large-@xmath36 region , not really a good solution but a local @xmath66 minimum , where the bathtub is much shallower and wider .
matter effects can also arise in the earth .
they are not msw ( no chance of adiabatic level crossing ) , just amplified vacuum oscillations through which @xmath31 can convert back to @xmath1 when the sun is below the horizon , giving day / night and summer / winter asymmetries in counting rates .
there are 2 conditions for big earth effects . + ( i ) near - resonant amplification in the earth ( @xmath67 large ) : @xmath68 assuming rock density @xmath69 .
+ ( ii ) matter oscillation wavelength ( @xmath70 ) less than earth diameter ( @xmath71 ) . at resonance
the matter - eigenvalue difference is @xmath72 , giving @xmath73 both these conditions can be approached or satisfied in a small region of @xmath74 for given @xmath75 , but @xmath36 can not be very small . for the kamiokande @xmath3 range , @xmath76 mev , a region near the msw large-@xmath36 solution is sensitive to earth effects ; the absence of a day / night asymmetry@xcite excludes this region ( labelled `` excluded 90% c.l . '' in fig.6 ) .
future experiments will enlarge this region of sensitivity .
similar things can happen for @xmath1 mixing with sterile @xmath39 , but now z - exchange no longer drops out ; coherent @xmath34 and @xmath77 z - exchanges cancel and the net effect is to replace @xmath37 above by @xmath78 where @xmath79 is the neutron number density@xcite .
the critical parameters change a bit and the large-@xmath36 solution vanishes ( fig.6b ) .
three- or four - flavour neutrino mixing offers more complicated possibilities , with more free parameters , that we do not need yet and shall not discuss today .
* 4 . exotic neutral current effects . *
if there are new neutral - current interactions , such as @xmath80 flavour - conserving or flavour - flipping scattering via r - parity - violating b - squark exchanges@xcite , new terms will appear in the matter - propagation matrix .
in the most general case with diagonal and off - diagonal contributions from scattering on @xmath81 distributions in matter , this matrix can be put in the form @xmath82 in the approximation @xmath83 , with just two constant parameters @xmath84 and @xmath85 describing the new physics in units of the standard matter effect . if @xmath86 , we have mixing and oscillations even in the absence of vacuum mixing ( @xmath87)@xcite .
these new terms modify the previous msw solutions .
fig.7 compares @xmath88 solutions in the cases @xmath89 , @xmath90 , @xmath91 ( with @xmath92)@xcite . adding this small exotic mixing
scarcely affects the large-@xmath36 solution but distorts or even splits the small-@xmath36 solution . *
5 . outlook .
* the different two - flavour - mixing scenarios can be distinguished ( or rejected ) by future measurements of the @xmath7 spectrum modulation ( sno , super - kamiokande , see figs.8,9 ) , the @xmath8 0.86 mev line contribution ( borexino ) , and possible day / night effects ( sno , super - kamiokande , icarus ) : @xmath93 furthermore , the charged - current / neutral - current event ratio [ sno , borex , icarus ] will distinguish whether the neutrino flavour mixed with @xmath1 is active ( @xmath94 ) or sterile ( @xmath39 ) .
the problem will become much more clearly defined .
99 j.m.bahcall , m.h.pinsonneault , rev .
mod . phys.*64 * , 885 ( 1992 ) .
s.turck-chieze , i.lopez , astrophys.j . * 408 * , 347 ( 1993 ) r.davis , jr .
, in frontiers of neutrino astrophysics ( eds . y.suzuki
, k. nakamura ) , universal academy press , tokyo 1993 .
k.s .
hirata et al , phys . rev .
* d44 * , 2241 ( 1991 ) ; a.suzuki , kek report 93 - 96 ; k.nakamura , talk at this conference .
sage collaboration : t.bowles , talk at this conference .
gallex collaboration , phys .
lett . *
b314 * , 445 ( 1993 ) and t.kirsten , talk at this conference . j.n .
bahcall , h.a .
bethe , phys . rev . *
d47 * , 1298 ( 1993 ) ; s.
bludman et al , ibid * d47 * , 2220 ( 1993 ) ; n.hata , p.langacker , ibid * d48 * , 2937 ( 1993 ) .
v.barger , r.j.n.phillips , k.whisnant , phys .
rev .
* d24 * , 538 ( 1981 ) , phys .
rev .
lett . *
69 * , 3135 ( 1992 ) ; s.l .
glashow , l.m .
krauss , phys .
lett .
* 190b * , 199 ( 1987 ) ; a. acker , s. pakvasa and j. pantaleone , phys . rev . *
d43 * , 1754 ( 1991 ) .
p.i.krastev , s.t.petkov , sissa preprint 177/93/ep .
sno collaboration : d.wark , talk at this conference .
r.s.raghavan , talk at this conference . a.yu.smirnov ,
d.n.spergell , j.n.bahcall , princeton preprint ast-93/15 .
l.wolfenstein , phys .
rev.*d17 * , 2369 ( 1978 ) ; * d20 * , 2634(1979 ) .
s.p .
mikheyev , a.yu .
smirnov , jetp * 91 * , 7 ( 1986 ) .
s.j .
parke , phys .
rev .
lett .
* 57 * , 1275 ( 1986 ) .
n.hata , p.langacker , pennsylvania preprint upr-0592-t ( nov.1993 ) .
v.barger et al , phys . rev . * d43 * , 1759 ( 1991 ) .
m.m.guzzo , a.masiero , s.t.petcov , phys . lett . *
b260 * , 154 ( 1991 ) ; e.roulet , phys.rev . * d44 * , 935 ( 1991 ) ; v.barger et al , phys . rev . *
d44 * , 1629 ( 1991 ) .
g.l .
fogli , e. lisi , preprint bari - th/135 - 93 .
|
we denote by @xmath3 the * binary boole algebra * , endowed with the discrete topology and with the usual laws . let be the boolean function @xmath4 and @xmath5 we define
@xmath6 by @xmath7@xmath8 @xmath9 represents the function resulting from @xmath10 when this one is not computed , in general , on all the coordinates @xmath11 if @xmath12
then @xmath13 is not computed , @xmath14 and if @xmath15 then @xmath13 is computed , @xmath16 let be the sequence @xmath17 the functions @xmath18 are defined iteratively by @xmath19@xmath20 the sequence @xmath21 is called * progressive * if@xmath22 the set of the progressive sequences is denoted by @xmath23 let be @xmath24 when @xmath25 is progressive , each coordinate @xmath26 is computed infinitely many times in the sequence @xmath27 @xmath28 .
this is the meaning of the progress property , giving the so called unbounded delay model of computation of the boolean functions .
the * initial value * , denoted by @xmath29 or @xmath30 and the * final value * , denoted by @xmath31 or @xmath32 of the function @xmath33 are defined by@xmath34@xmath35 the function @xmath33 is called ( * pseudo*)*periodical with the period * @xmath36 if \a ) @xmath37 does not exist and \b ) @xmath38 the * characteristic function * @xmath39 of the set @xmath40 is defined in the following way:@xmath41{c}1,\;if\;t\in a\;\\ 0,otherwise \end{array } \right . .\ ] ] we denote by @xmath42 the set of the real sequences @xmath43 which are unbounded from above . the sequences @xmath44 act as time sets . at this level of generality of the exposure
, a double uncertainty exists in the real time iterative computations of the function @xmath45 we do not know precisely neither the coordinates @xmath13 of @xmath10 that are computed , nor when the computation happens .
this uncertainty implies the non determinism of the model and its origin consists in structural fluctuations in the fabrication process , the variations in ambiental temperature and the power supply etc . a * signal * ( or @xmath46**signal * * ) is a function @xmath33 of the form@xmath47@xmath48 with @xmath49 the set of the signals is denoted by @xmath50 the signals @xmath51 model the electrical signals from the digital electrical engineering .
they have by definition initial values and they avoid dirichlet type properties ( called zeno properties by the engineers ) such as@xmath52@xmath53 because these properties can not characterize the inertial devices .
we denote by @xmath54 the set of the non - empty subsets of a set .
the * autonomous asynchronous systems * are the non - empty sets @xmath55 [ exa13]we give the following simple example that shows how the autonomous asynchronous systems model the asynchronous circuits . in figure [ preliminaries1 ]
we have drawn the ( logical ) gate not with the [ ptb ] input @xmath56 and the state ( the output ) @xmath57 for @xmath58 and@xmath59 the state @xmath60 represents the computation of the negation of @xmath61 and it is of the form@xmath62@xmath63 where @xmath64 is the initial value of @xmath60 and @xmath44 is arbitrary .
as we can see , @xmath60 depends on @xmath65 only and it is independent on @xmath66 in figure [ preliminaries2 ] , [ ptb ] we have@xmath67@xmath68 thus this circuit is modeled by the autonomous asynchronous system@xmath69@xmath70 the * progressive functions * @xmath71 are by definition the functions@xmath72 where @xmath44 and @xmath73 the set of the progressive functions is denoted by @xmath74 [ def20]for @xmath75 and @xmath76 like at ( [ pre2 ] ) , we define @xmath77 by @xmath78@xmath79 the previous equation reminds the iterations of a discrete time real dynamical system . the time is not exactly discrete in it , but some sort of intermediate situation occurs between the discrete and the real time ; on the other hand the iterations of @xmath10 do not happen on all the coordinates ( synchronicity ) , but on some coordinates only , such that any coordinate @xmath13 is computed infinitely many times , @xmath80 ( asynchronicity ) when @xmath81 .
we denote by@xmath82 the discrete time set . let be @xmath75 and @xmath83 we define the function @xmath84 by @xmath85@xmath86{c}\mu , k=-1,\\ \phi^{\alpha^{0} ... \alpha^{k}}(\mu),k\geq0 \end{array } \right . .\ ] ] let us denote@xmath87 the equivalence of @xmath88 is defined by : @xmath89 such that ( [ pre2 ] ) and@xmath90 are true .
the canonical surjection @xmath91 is by definition the function @xmath92@xmath93 where @xmath94 is the only sequence such that @xmath95 exists , making the equation ( [ pre2 ] ) true .
the relation between the continuous and the discrete time is the following : for any @xmath96 and any @xmath97 @xmath98 and @xmath44 exist making the equation ( [ pre2 ] ) true and we have@xmath99@xmath100 equivalent progressive functions @xmath88
( i.e. @xmath101 ) give equivalent functions @xmath102 in the sense that the computations of @xmath10 are the same , but the time flow is piecewise faster or slower in the two situations .
the * universal regular autonomous asynchronous system * @xmath103 that is generated by the function @xmath75 is defined by @xmath104 an autonomous asynchronous system @xmath105 is called * regular * , if @xmath10 exists such that @xmath106 in this case @xmath10 is called the * generator function network function in a non - autonomous , discrete time context ; for moisil , network means system or circuit. ] * of @xmath107 . in the last two definitions ,
the attribute regular refers to the existence of a generator function @xmath10 and the attribute universal means maximal relative to the inclusion . for a regular system ,
@xmath10 is not unique in general . for any @xmath108 and @xmath109 the autonomous systems @xmath110 @xmath111 @xmath112 and @xmath113 are regular . for @xmath114
the system @xmath115 the constant function , for which @xmath116 .
these examples suggest several possibilities of defining the systems @xmath117 which are not universal .
for example by putting appropriate supplementary requests on the functions @xmath118 one could rediscover the bounded delay model of computation of the boolean functions .
let be @xmath119 two things are understood by * orbit * , or ( * state * , or * phase * ) * trajectory * ( @xcite , page 19 ; @xcite , page 3 ; @xcite , page 8 ; @xcite , page 24 ; @xcite , page 2 ) * of * @xmath113 * starting at * @xmath96 : \a ) the function @xmath120 \b ) the set @xmath121 representing the values of the previous function . sometimes ( @xcite , page 4 ; @xcite , page 91 ; @xcite , page 24 ; @xcite , page 2 ) the function from a ) is called the * motion * ( or the * dynamic * ) of @xmath122 through @xmath123 the equivalent properties @xmath124 and @xmath125 are called of * accessibility * ; the points @xmath126 are said to be * accessible*. [ rem32]the orbits are the curves in @xmath127 , parametrized by @xmath128 and @xmath129 on the other hand @xmath97 @xmath130 imply @xmath131 and we see the truth of the implication@xmath132 the * state * ( or the * phase * ) * portrait * of @xmath113 is the set of its orbits ( @xcite , page 4 ; @xcite , page 92 ; @xcite , page 10 ; @xcite , page 2 ) .
[ exa34]the function @xmath133 is defined by the following table@xmath134{cc}(\mu_{1},\mu_{2 } ) & \phi(\mu_{1},\mu_{2})\\ ( 0,0 ) & ( 0,0)\\ ( 0,1 ) & ( 1,0)\\ ( 1,0 ) & ( 1,1)\\ ( 1,1 ) & ( 1,1 ) \end{array}\ ] ] the state portrait of @xmath113 is:@xmath135@xmath136@xmath137@xmath138 this set is drawn in figure [ ph1 ] , [ ptb ] where the arrows show the increase of time .
one might want to put arrows from @xmath139 to itself and from @xmath140 to itself .
let be @xmath141 for any @xmath142 the * nullclins * of @xmath10 are the sets@xmath143 if @xmath144 then the coordinate @xmath145 is said to be * not excited * , or * not enabled * , or * stable * and if @xmath146 then it is called * excited * , or * enabled * , or * unstable*. sometimes , instead of indicating @xmath10 by a table like previously , we can replace figure [ ph1 ] by figure [ ph2 ] , [ ptb ] where we have underlined the unstable coordinates . for example in figure [ ph2 ] , @xmath147 means that @xmath148 @xmath149 means that @xmath150 etc .
in fact figure [ ph2 ] results uniquely from figure [ ph1 ] , one could know by looking at figure [ ph1 ] which coordinates should be underlined and which should be not .
a point @xmath96 that fulfills @xmath151 is called a * fixed point * ( an * equilibrium point * , a * critical point * , a * singular point * ) ( @xcite , page 43 ; @xcite , page 4 ; @xcite , page 92 ; @xcite , page 9 ; @xcite , page 24 ; @xcite , page 2 ) , shortly an * equilibrium * of @xmath152 a point that is not fixed is called * ordinary*. [ the38]the following statements are equivalent for @xmath153@xmath154@xmath155@xmath156@xmath157@xmath158@xmath159 ( [ equ1])@xmath160([equ2 ] ) we take @xmath161 in the following way@xmath162 with @xmath49 for the sequence@xmath163 from @xmath164 we can prove by induction on @xmath165 that@xmath166 wherefrom@xmath167 ( [ equ2])@xmath160([equ1 ] ) from ( [ equ2 ] ) we have the existence of @xmath168 and @xmath44 with the property that ( [ equ7 ] ) is true , thus ( [ equ6 ] ) is true .
we denote@xmath169@xmath170@xmath171@xmath172@xmath171 and we have from ( [ equ6 ] ) : @xmath173@xmath174{c}\phi_{i}(\mu),i\in i_{0}\\ \mu_{i},i\in\{1, ... ,n\}\setminus i_{0}\end{array } \right . = \mu_{i};\ ] ] @xmath175@xmath176{c}\phi_{i}(\mu),i\in i_{1}\\ \mu_{i},i\in\{1, ...
,n\}\setminus i_{1}\end{array } \right . = \mu_{i};\]]@xmath171 @xmath177@xmath178{c}\phi_{i}(\mu),i\in i_{k}\\ \mu_{i},i\in\{1, ... ,n\}\setminus i_{k}\end{array } \right . = \mu_{i};\]]@xmath171 with the conclusion that@xmath179 some @xmath180 exists with the property that@xmath181 thus ( [ equ1 ] ) is true .
( [ equ1])@xmath160([equ2 _ ] ) let be@xmath182 with @xmath183 and @xmath44 arbitrary .
it is proved by induction on @xmath165 the validity of ( [ equ6 ] ) and this implies the truth of ( [ equ7 ] ) .
( [ equ2_])@xmath160([equ1 ] ) this is true because ( [ equ2_])@xmath160([equ2 ] ) and ( [ equ2])@xmath160([equ1 ] ) are true .
( [ equ2])@xmath184([equ3 ] ) and ( [ equ2_])@xmath184([equ3 _ ] ) are obvious .
( [ equ1])@xmath184([equ3 _ _ ] ) @xmath185 and ... and @xmath186 and ... and @xmath187 if @xmath188 then @xmath92 the orbit @xmath189 is called * rest position*.
[ the8](@xcite , theorem 49 ) the following fixed point property is true@xmath190 let @xmath191 be arbitrary and fixed .
some @xmath130 exists such that @xmath192@xmath193 and from theorem [ the38 ] , ( [ equ2])@xmath160([equ1 ] ) we have @xmath194 theorem [ the8 ] shows that the final values of the states of a system are fixed points of @xmath10 .
[ the9](@xcite , theorem 50 ) we have @xmath195@xmath196 for arbitrary @xmath197 we suppose that @xmath198and@xmath199 we have @xmath200@xmath201 as resulting from theorem [ the9 ] , the accessible fixed points are final values of the states of the systems .
the properties of the fixed points that are expressed by theorems [ the38 ] , [ the8 ] , [ the9 ] give a better understanding of example [ exa34 ] .
[ def108]the system @xmath113 ( or @xmath10 ) is * transitive * ( @xcite , page 22 ; @xcite , page 3 ) , or * minimal * ( @xcite , page 23 ) if one of the following non - equivalent properties holds true:@xmath202@xmath203 the property of transitivity may be considered one of surjectivity or one of accessibility .
if @xmath10 is transitive , then it has no fixed points .
the property ( [ tt1 ] ) of transitivity is exemplified in figure [ minimality1 ] [ ptb ] and the property ( [ tt2 ] ) of transitivity is exemplified in figure [ minimality2 ] .
let @xmath204 and @xmath33 be some functions .
we denote by @xmath205 the function@xmath206 if @xmath204 and @xmath51 is expressed by@xmath207 then@xmath208@xmath209 for @xmath204 and @xmath210 we denote by @xmath211 the sequence @xmath212 let be @xmath213 arbitrary and we denote for @xmath214@xmath215 we denote by @xmath216 the set of the functions @xmath204 that fulfill \i ) @xmath217 is bijective ; \ii ) @xmath218
\iii ) @xmath219@xmath220 [ the59]a ) @xmath216 is group relative to the composition @xmath221 of the functions ; \b ) @xmath222 \c ) @xmath223 \a ) the fact that @xmath224 @xmath225 and @xmath226 is obvious .
\b ) let @xmath227 and @xmath228 be arbitrary .
we denote for @xmath229@xmath230 and we remark that@xmath231@xmath232{c}\mu^{1}=(1, ...
,1),p=1\\ \mu^{1}\cup ...
\cup\mu^{p}=(1, ... ,1),p\geq2 \end{array } \right . , \]]@xmath233@xmath232{c}h(\mu^{1})=(1, ...
,1),p=1\\ h(\mu^{1})\cup ...
\cup h(\mu^{p})=(1, ...
,1),p\geq2 \end{array } \right . .\ ] ] case @xmath234@xmath235 case @xmath236@xmath237@xmath238 \c ) let us take arbitrarily some @xmath227 and a function @xmath239@xmath240 where @xmath168 and @xmath49 we have@xmath241@xmath242@xmath243@xmath244@xmath245@xmath246 because @xmath247 taking into account b ) , we conclude that @xmath248 [ the60]let be the generator functions @xmath249 of the systems @xmath250 and the bijections @xmath251 the following statements are equivalent : \a ) @xmath252 the diagrams@xmath134{ccc}\mathbf{b}^{n } & \overset{\phi^{\nu}}{\rightarrow } & \mathbf{b}^{n}\\ h\downarrow\ ; & & \;\downarrow h\\ \mathbf{b}^{n } & \overset{\psi^{h^{\prime}(\nu)}}{\rightarrow } & \mathbf{b}^{n}\end{array}\ ] ] are commutative ; \b ) @xmath253@xmath254 \c ) @xmath255@xmath256 a)@xmath160b ) it is sufficient to prove that @xmath257@xmath258 since this is equivalent with b ) .
we fix arbitrarily some @xmath122 and some @xmath259 and we use the induction on @xmath165 . for @xmath260 the statement is proved , thus we suppose that it is true for @xmath165 and we prove it for @xmath261:@xmath262@xmath263 b)@xmath160c ) for arbitrary @xmath96 and @xmath239@xmath264 @xmath265 we have that@xmath266 is an element of @xmath267 ( see theorem [ the59 ] c ) ) and@xmath268@xmath269@xmath270@xmath271@xmath272@xmath273 c)@xmath160a ) let @xmath274 be arbitrary and fixed and we consider @xmath97@xmath275 with @xmath44 fixed too .
we have@xmath276@xmath277 but@xmath278@xmath279@xmath280 and from ( [ eds1 ] ) , for @xmath281 we obtain@xmath282 [ def160]we consider the generator functions @xmath249 and the universal asynchronous systems @xmath283 @xmath284 .
if two bijections @xmath285 exist such that one of the equivalent properties a ) , b ) , c ) from theorem [ the60 ] is satisfied , then @xmath286 are called * equivalent * ( @xcite , page 35 ; @xcite , page 102 ; @xcite , page 40 ; @xcite , page 32 ; @xcite , page 6 ) and @xmath287 are called * conjugated*. in this case we denote @xmath288 [ def161]we fix @xmath152 the fact that @xmath289 exists such that the previous property holds , makes us say that @xmath10 is * structurally stable * ( peixoto @xcite , page 121 ) .
@xmath290 is called an * admissible * ( or * allowable * ) * perturbation of * @xmath10 .
the equivalence of the universal regular autonomous asynchronous systems is indeed an equivalence and it should be understood as a change of coordinates . thus @xmath10 and @xmath290 are indistinguishable .
@xmath291 are given by , see figure [ echiv1 ] [ ptb ] @xmath292@xmath293 and the bijection @xmath294 is@xmath295 the diagram@xmath134{ccc}\mathbf{b}^{2 } & \overset{\phi^{\nu}}{\rightarrow } & \mathbf{b}^{2}\\ h\downarrow\ ; & & \;\downarrow h\\ \mathbf{b}^{2 } & \overset{\psi^{\nu^{\prime}}}{\rightarrow } & \mathbf{b}^{2}\end{array}\ ] ] commutes for @xmath296 and for @xmath297 we have the assignments@xmath134{ccc}(0,0 ) & \overset{\phi}{\rightarrow } & ( 0,1)\\ h\downarrow\ ; & & \;\downarrow h\\ ( 1,1 ) & \overset{\psi}{\rightarrow } & ( 0,1 ) \end{array } , \;\begin{array } [ c]{ccc}(0,1 ) & \overset{\phi}{\rightarrow } & ( 1,0)\\ h\downarrow\ ; & & \;\downarrow h\\ ( 0,1 ) & \overset{\psi}{\rightarrow } & ( 1,0 ) \end{array } , \;\begin{array } [ c]{ccc}(1,0 ) & \overset{\phi}{\rightarrow } & ( 1,1)\\ h\downarrow\ ; & & \;\downarrow h\\ ( 1,0 ) & \overset{\psi}{\rightarrow } & ( 0,0 ) \end{array } , \;\begin{array } [ c]{ccc}(1,1 ) & \overset{\phi}{\rightarrow } & ( 0,0)\\ h\downarrow\ ; & & \;\downarrow h\\ ( 0,0 ) & \overset{\psi}{\rightarrow } & ( 1,1 ) \end{array } .\ ] ] we denote @xmath298@xmath299 for @xmath300 we have@xmath134{ccc}(0,0 ) & \overset{(\pi_{1},\phi_{2})}{\rightarrow } & ( 0,1)\\ h\downarrow\ ; & & \;\downarrow h\\ ( 1,1 ) & \overset{(\psi_{1},\pi_{2})}{\rightarrow } & ( 0,1 ) \end{array } , \begin{array } [ c]{ccc}(0,1 ) & \overset{(\pi_{1},\phi_{2})}{\rightarrow } & ( 0,0)\\ h\downarrow\ ; & & \;\downarrow h\\ ( 0,1 ) & \overset{(\psi_{1},\pi_{2})}{\rightarrow } & ( 1,1 ) \end{array } , \begin{array } [ c]{ccc}(1,0 ) & \overset{(\pi_{1},\phi_{2})}{\rightarrow } & ( 1,1)\\ h\downarrow\ ; & & \;\downarrow h\\ ( 1,0 ) & \overset{(\psi_{1},\pi_{2})}{\rightarrow } & ( 0,0 ) \end{array } , \begin{array } [ c]{ccc}(1,1 ) & \overset{(\pi_{1},\phi_{2})}{\rightarrow } & ( 1,0)\\ h\downarrow\ ; & & \;\downarrow h\\ ( 0,0 ) & \overset{(\psi_{1},\pi_{2})}{\rightarrow } & ( 1,0 ) \end{array}\ ] ] and for @xmath301 the assignments are@xmath134{ccc}(0,0 ) & \overset{(\phi_{1},\pi_{2})}{\rightarrow } & ( 0,0)\\ h\downarrow\ ; & & \;\downarrow h\\ ( 1,1 ) & \overset{(\pi_{1},\psi_{2})}{\rightarrow } & ( 1,1 ) \end{array } , \begin{array } [ c]{ccc}(0,1 ) & \overset{(\phi_{1},\pi_{2})}{\rightarrow } & ( 1,1)\\ h\downarrow\ ; & & \;\downarrow h\\ ( 0,1 ) & \overset{(\pi_{1},\psi_{2})}{\rightarrow } & ( 0,0 ) \end{array } , \begin{array } [ c]{ccc}(1,0 ) & \overset{(\phi_{1},\pi_{2})}{\rightarrow } & ( 1,0)\\ h\downarrow\ ; & & \;\downarrow h\\ ( 1,0 ) & \overset{(\pi_{1},\psi_{2})}{\rightarrow } & ( 1,0 ) \end{array } , \begin{array } [ c]{ccc}(1,1 ) & \overset{(\phi_{1},\pi_{2})}{\rightarrow } & ( 0,1)\\ h\downarrow\ ; & & \;\downarrow h\\ ( 0,0 ) & \overset{(\pi_{1},\psi_{2})}{\rightarrow } & ( 0,1 ) \end{array}\ ] ] respectively . @xmath10 and
@xmath290 are conjugated .
the functions @xmath302 are given in the following table@xmath134{ccc}(\mu_{1},\mu_{2 } ) & h(\mu_{1},\mu_{2 } ) & h^{\prime}(\mu_{1},\mu_{2})\\ ( 0,0 ) & ( 0,1 ) & ( 0,0)\\ ( 0,1 ) & ( 1,1 ) & ( 1,0)\\ ( 1,0 ) & ( 0,0 ) & ( 0,1)\\ ( 1,1 ) & ( 1,0 ) & ( 1,1 ) \end{array}\ ] ] and the state portraits of the two systems are given in figure [ echiv2 ] .
@xmath113 and @xmath284 are equivalent .
[ ptb ] if @xmath10 and @xmath290 are conjugated , then the following possibilities exist : \a ) @xmath303 \b ) @xmath304 and @xmath305 we presume that @xmath288 in the equation@xmath306 we put @xmath307 and we have@xmath308 thus @xmath309 and finally @xmath310 we suppose that @xmath113 and @xmath284 are equivalent and let be @xmath311 such that @xmath312 .
\a ) if @xmath122 is a fixed point of @xmath313 then @xmath314 is a fixed point of @xmath315
\b ) for any @xmath96 and any @xmath97 if @xmath316 is periodical with the period @xmath317 , then @xmath318 is periodical with the period @xmath317 .
\c ) if @xmath113 is transitive , then @xmath284 is transitive .
\a ) the commutativity of the diagram@xmath134{ccc}\mathbf{b}^{n } & \overset{\phi^{\nu}}{\rightarrow } & \mathbf{b}^{n}\\ h\downarrow\ ; & & \;\downarrow h\\ \mathbf{b}^{n } & \overset{\psi^{h^{\prime}(\nu)}}{\rightarrow } & \mathbf{b}^{n}\end{array}\ ] ] for @xmath319 gives@xmath320@xmath321 \b ) the hypothesis states that @xmath322@xmath323 and in this situation@xmath324 \c ) let @xmath325 be arbitrary and fixed .
the hypothesis ( [ tt1 ] ) states that@xmath326 wherefrom@xmath327 the situation with ( [ tt2 ] ) is similar .
let be the generator function @xmath328 @xmath329 that depends on the parameter @xmath330 . intuitively speaking ( ott , @xcite , page 137 ) a dynamic bifurcation is a qualitative change in the dynamic of the system @xmath331 that occurs at the variation of the parameter @xmath332 .
if for any parameters @xmath333 the systems @xmath331 and @xmath334 are equivalent , then @xmath10 is called * structurally stable * ( @xcite , page 117 ; @xcite , page 43 ; @xcite , page 9 ) ; the existence of @xmath335 such that @xmath331 and @xmath334 are not equivalent is called a * dynamic * * bifurcation * ( @xcite , page 57 ; @xcite , page 9 ) .
equivalently , let us fix an arbitrary @xmath336 if @xmath337 , @xmath338 is an admissible perturbation of @xmath339 ( definition [ def161 ] ) , then @xmath10 is said to be * structurally stable * , otherwise we say that @xmath10 has a * dynamic bifurcation*. if @xmath340 the bijections @xmath341 exist such that @xmath252 the diagram@xmath134{ccc}\;\;\mathbf{b}^{n } & \overset{\phi^{\nu}(\cdot,\lambda)}{\rightarrow } & \mathbf{b}^{n}\;\\ h\downarrow & & \downarrow h\\ \;\;\mathbf{b}^{n } & \overset{\phi^{h^{\prime}(\nu)}(\cdot,\lambda^{\prime})}{\rightarrow } & \mathbf{b}^{n}\ ; \end{array}\ ] ] commutes , then @xmath10 is structurally stable , otherwise we have a dynamic bifurcation . @xmath10 is structurally stable and the bijections @xmath311 are defined accordingly to the following table:@xmath134{ccc}(\mu_{1},\mu_{2 } ) & h(\mu_{1},\mu_{2 } ) & h^{\prime}(\mu_{1},\mu_{2})\\ ( 0,0 ) & ( 0,1 ) & ( 0,0)\\ ( 0,1 ) & ( 1,1 ) & ( 1,0)\\ ( 1,0 ) & ( 0,0 ) & ( 0,1)\\ ( 1,1 ) & ( 1,0 ) & ( 1,1 ) \end{array}\ ] ] the * bifurcation diagram * ( @xcite , page 61 ) is a partition of the set of systems @xmath343 in classes of equivalence given by the equivalence of the systems , together with representative state portraits for each class of equivalence .
let be @xmath344 .
the families of systems @xmath345 and @xmath346 are called * equivalent * ( @xcite , pages 7 , 17 ) if there exists a bijection @xmath347 such that @xmath348 and @xmath349 are equivalent in the sense of definition [ def160 ] .
|
the asynchronous systems are the non - deterministic models of the asynchronous circuits from the digital electrical engineering . in the autonomous version ,
such a system is a set of functions @xmath0 called states ( @xmath1 is the time set ) .
if an asynchronous system is defined by making use of a so called generator function @xmath2 then it is called regular .
the property of universality means the greatest in the sense of the inclusion .
the purpose of the paper is that of defining and of characterizing the fixed points , the equivalencies and the dynamical bifurcations of the universal regular autonomous asynchronous systems .
we use analogies with the dynamical systems theory .
|
we present a phenomenological model that seeks to explain abundances of a large number of elements in ultra - metal - poor ( ump ) stars with [ fe / h ] @xmath9 and extend this model to metal - poor ( mp ) stars with @xmath8}<-1 $ ] . recently
burris et al . ( 2000 ) reported abundances of heavy elements , both above and below ba , in 70 galactic halo stars with a wide range in [ fe / h ] .
this new work and previous observations by e.g. , mcwilliam et al .
( 1995 ) and ryan , norris , & beers ( 1996 ) stimulated us to address the problem of chemical evolution in the early galaxy .
recent studies have discussed galactic chemical evolution of h to zn ( timmes , woosley , & weaver 1995 ) and of heavy elements from ba to eu with both @xmath10-process and @xmath0-process contributions ( travaglio et al .
these studies have shown general trends for abundance ratios of different elements to fe as a function of [ fe / h ] that converge on solar values .
raiteri et al .
( 1999 ) treated temporal evolution of asymptotic giant branch ( agb ) stars and rates of type ii supernovae ( snii ) and simulated galactic ba enrichment .
their results on ba / fe show a wide scatter at low [ fe / h ] as observed .
this was attributed to local inhomogeneities in the interstellar medium ( ism ) . in the present work
, we will show that the abundances in ump and mp stars follow almost quantitatively from a set of simple rules .
the present study is purely phenomenological and confined to @xmath11}<-1 $ ] where type ia supernovae ( snia , an fe source ) and low mass agb stars ( the dominant @xmath10-process source ) would not contribute significantly to the ism . as pointed out by truran ( 1981 ) , abundances of neutron - capture elements ( e.g. , ba ) at low [ fe / h ] are dominated by @xmath0-process contributions from snii .
we focus on the most extensive and self - consistent data set so far available , that of burris et al .
( 2000 ) , but treat data sets of magain ( 1989 ) , gratton & sneden ( 1994 ) , and johnson & bolte ( 2001 ) as well .
we will present the abundances resulting from production by the first generations of stars formed after the big bang and yields for two hypothesized types of snii .
these results will be inferred from the observational data and not from theoretical models for the first generations of stars or for @xmath0-process production by snii .
the @xmath0-process abundances ( @xmath0-abundances ) calculated from our model for ump and mp stars will be compared with the observational data .
more specifically , the @xmath0-abundance of an element that represents the total abundance of all @xmath0-process isotopes of the element is calculated for comparison with observations as there are no data on isotopic @xmath0-abundances in stars other than the sun .
when referring to @xmath0-process nuclei , we consider both those that are the direct products of a rapid neutron capture scenario and those that may be produced by other processes closely related to the @xmath0-process in the same snii event .
the latter would include many nuclei with atomic masses of @xmath12 that would be made in the @xmath13-process ( woosley & hoffman 1992 ) .
if a sufficient neutron abundance existed at the end of the @xmath13-process , the nuclei with @xmath12 would become the seed nuclei to capture the neutrons during the subsequent @xmath0-process .
otherwise , these nuclei would experience no further processing and be a part of the snii ejecta .
the @xmath13-process could be the dominant source of sr , y and zr .
it is assumed here that snii are the source of @xmath0-process nuclei ( @xmath0-nuclei ) .
the abundances of the @xmath0-nuclei @xmath14hf and @xmath15i in the early solar system are well established .
these data require that the snii responsible for @xmath14hf not produce any significant amounts of @xmath15i [ wasserburg , busso , & gallino 1996 ( wbg ) ] .
the frequency of this type of snii was argued to be @xmath16 for a standard reference mass of hydrogen ( see [ model2 ] ) .
the nuclide @xmath15i was attributed to a different type of snii with a frequency of @xmath17 .
these hypothetical snii types are called @xmath3 and @xmath4 events for the `` high '' and `` low '' frequency types , respectively .
this approach assigns the dominant production of nuclei with @xmath6 ( `` heavy '' @xmath0-nuclei ) to the @xmath3 events and the nuclei with @xmath7 ( `` light '' @xmath0-nuclei ) to the @xmath4 events .
qian , vogel , & wasserburg ( 1998 ) developed a simple @xmath0-process model involving neutrinos to examine the different conditions in @xmath3 and @xmath4 events that are needed to provide a split between the production of heavy and light @xmath0-nuclei at @xmath18 . due to their much higher frequency
, the @xmath3 events were expected to be likely the first snii to inject @xmath0-nuclei into the ism ( wbg ) .
the effects of such first generation " snii should be apparent in the abundances of @xmath0-elements in ump stars .
exquisite observations by sneden et al .
( 1996 , 2000 ) and westin et al .
( 2000 ) have established that abundance ratios of many other heavy @xmath0-elements relative to eu in ump stars are remarkably constant and close to solar @xmath0-process values .
this demonstrates that abundances of heavy @xmath0-elements in ump stars appear to reflect production by pure @xmath3 events .
sneden et al .
( 2000 ) also found that the abundances of light @xmath0-elements such as pd , ag , and cd in the ump star cs 22892 - 052 were low relative to the solar @xmath0-abundance pattern that is translated to pass through the eu data .
this provides additional evidence that there should be at least two distinct types of snii sources for the @xmath0-nuclei . the available data on ump stars ( e.g. , mcwilliam et al .
1995 ; mcwilliam 1998 ; westin et al . 2000 ; sneden et al .
2000 ; burris et al . 2000 ) show that there is a wide range in the abundances of the heavy @xmath0-elements eu and ba ( @xmath19 dex ) for stars with @xmath11}<-2.5 $ ] .
this clearly indicates that heavy @xmath0-elements , including the chronometer th , are produced without any significant coproduction of fe , thus requiring that heavy @xmath0-elements and fe be produced in different types of snii .
there is little ba in stars with @xmath20}<-3 $ ] ( mcwilliam et al .
1995 ; mcwilliam 1998 ) .
wasserburg & qian [ 2000 ( wq ) ] inferred that the large dispersion in eu and ba abundances at [ fe / h]@xmath21 was caused by @xmath3 events adding heavy @xmath0-elements but no fe to a pre - existing inventory of elements in the ism .
this pre - existing inventory of fe and associated elements was called the initial or prompt ( hereafter @xmath22 ) inventory and attributed to the first generations of very massive ( @xmath23 ) stars formed from big bang debris .
it was further inferred that normal stars with masses of @xmath24@xmath25 could only form in a medium with sufficient metals to permit cooling during the collapse of a gas cloud .
normal stars with masses of @xmath26@xmath25 would later become snii .
further evidence for the decoupling of heavy @xmath0-elements from fe has been provided by the observations of the u - rich star cs 31082 - 001 ( cayrel
et al . 2001 ) .
this star with [ fe / h ] @xmath27 has extremely abundant os , ir , th , and u , sharply exhibiting the effect that little fe is coproduced with the heavy @xmath0-elements as argued by qian & wasserburg [ 2001b ( qw ) ] .
figure [ ba ] shows the available ba data over @xmath20}\lesssim -1 $ ] and includes the ba abundance inferred by qw for cs 31082 - 001 based on the three - component ( @xmath22 , @xmath3 , and @xmath4 ) model for abundances in ump and mp stars .
this inferred abundance ( upper cross ) is within @xmath28 dex of the preliminary ba data on cs 31082 - 001 ( lower cross ) reported by hill et al .
( 2001 ) . as discussed above , the large scatter in ba abundance at @xmath29}<-2.5 $ ] shown in figure [ ba ]
is attributed to the rapid occurrence of @xmath3 events shortly after the onset of normal star formation .
the scenario adopted here for the evolution of @xmath0-process related elements over galactic history is schematically shown in table [ epoch ] .
some time after the big bang , baryonic matter condenses in dark matter potential wells .
only very massive stars can form from big bang debris .
these stars explode and contribute newly - synthesized material . the number of such events is not known but many such disrupting events may be required to give [ fe / h ] @xmath9 in an average parcel of the ism or igm ( intergalactic medium ) .
the ism or igm now has a changed chemical composition compared with big bang debris .
when some of this material cools , there is condensation of gas to form a more or less normal stellar population that includes the progenitors for snii @xmath3 and @xmath4 events and lower mass stars .
this state of condensation can be achieved when the abundance of all metals in the ism or igm reaches or surpasses some critical value .
a nominal value of [ fe / h ] @xmath9 for the critical `` metallicity '' was inferred by wq from the observational data on ump stars .
this appears to be supported by the recent results of bromm et al .
( 2001 ) who have simulated the collapse and fragmentation of gas clouds with different metallicities .
the nominal value [ fe / h ] @xmath9 will be used below for the @xmath22-inventory of fe although the onset of `` normal '' astration may correspond to a possible range of @xmath30 } < -2.7 $ ] .
the first snii to contribute @xmath0-elements to the ism are the most frequent and most probable @xmath3 events . on a timescale of @xmath31 yr
, the relative contributions of @xmath3 and @xmath4 events should approach well - defined values that are determined by the average frequency ratio of @xmath32 and the yields of these events . from then on the evolution of @xmath0-elements and fe
would progress as if there were only one type of snii with small statistical deviations ( see the evolution of ba abundance relative to fe at [ fe / h ] @xmath33 indicated by the dot - dashed line in figure [ ba ] ) . at [ fe / h ]
@xmath34 , snia begin to contribute and dominate the fe production , as inferred from the o data ( e.g. , timmes et al . 1995 ; qian & wasserburg 2001a ) .
this approach sharply separates heavy element production in the earliest epochs [ ( 2 ) and ( 3 ) in table [ epoch ] ] from later production by snii [ ( 6 ) and ( 7 ) ] and from even later production by snia and contributions of evolved normal low - mass stars .
the numbers of @xmath3 and especially @xmath4 events through epochs ( 6 ) and ( 7 ) are not very large ( see [ nhl ] ) so that a discrete model is required . for simplicity it is assumed that each @xmath3 or @xmath4 event has a fixed relative abundance pattern for the elements produced in each case and that the yields are constant for each @xmath3 or @xmath4 event .
the average frequencies of these events are also assumed to be constant for a standard reference mass of hydrogen .
the main results of the model only depend on the assumption of fixed yield patterns for @xmath3 and @xmath4 events ( see [ model2][nhl ] ) .
the key parameters of the model that must be determined by consideration of the observational data are the yields of @xmath3 and @xmath4 events for pertinent elements and the @xmath22-inventory composition that defines the baseline to which the contributions of @xmath3 and @xmath4 events are added .
as we are concerned with the addition of elements to the ism or igm whose hydrogen mass fraction is essentially not altered during galactic chemical evolution , we will typically use a representation where the abundances are given relative to hydrogen .
this is the standard spectroscopic notation @xmath35 for element e where e is also used to represent the number of e atoms .
the common notation [ e / fe ] complicates the discussion and is avoided here as ranges in [ e / fe ] may result from addition of either element e or fe .
consider a homogeneous system of gas with a time - dependent total number h of hydrogen atoms .
the rate of change in the total number e of e atoms in this gas may be written as @xmath36 where @xmath37 is the total rate for injection of element e into the gas after production by its sources , and the last term accounts for the removal of matter from the gas by either astration or fragmentation ( @xmath38 for both cases ) .
the above equation may be rewritten as @xmath39 thus , if snii @xmath3 and @xmath4 events are the only sources under consideration for element e and the snii frequencies are proportional to the amount of hydrogen in the gas , then equation ( [ dehdt ] ) reduces to the following for a homogeneous mass of gas : @xmath40 or in the spectroscopic notation , @xmath41 in equation ( [ eh ] ) , ( e / h)@xmath42 is the @xmath22-inventory , ( e / h)@xmath43 or ( e / h)@xmath44 is the number of e atoms produced per hydrogen atom in the gas phase by each @xmath3 or @xmath4 event for a standard reference mass of hydrogen that is assumed here to mix with the ejecta in each case , and @xmath45 or @xmath46 is the number of @xmath3 or @xmath4 events that have occurred in this reference mass . now consider the mixture of two different gas systems ( with subscripts `` 1 '' and `` 2 '' ) that have different nucleosynthetic histories . for fixed values of ( e / h)@xmath42 , ( e / h)@xmath43 , and ( e / h)@xmath44 , the abundance of element e in the mixture of two gas masses
can be written as @xmath47 where @xmath48 it can be seen that a mixture of two gas systems with different nucleosynthetic histories will result in a new system with effective values of @xmath49 and @xmath50 corresponding to the weighted averages of the contributing systems . as a result ,
all that mixing will do is change the values of @xmath45 and @xmath46 but still produce abundance patterns congruent to those resulting from @xmath3 and @xmath4 events .
if the absolute yields of @xmath3 and @xmath4 events were to shift but the relative yields of one element to the other be maintained in each type of event , then the only result of this shift would again be to change @xmath45 and @xmath46 but the same congruence of abundance patterns would be maintained .
it follows that a wide range of removal by astration and mixing of matter with different chemical evolution lead to a result that is indistinguishable from a simple history if the @xmath22-inventory composition and the basic relative yield templates of @xmath3 and @xmath4 events are fixed as assumed in this model .
thus complex mixing and transport models are not required to pursue the approach used here but are effectively represented by the @xmath45 and @xmath46 values for a star .
if the model assumptions are correct , then they provide a full basis for quantitatively determining the @xmath0-abundances in ump and mp stars .
the frequencies of @xmath3 and @xmath4 events inferred from the abundances of @xmath14hf and @xmath15i in the early solar system , @xmath51 and @xmath52 , respectively , correspond to a reference mass of hydrogen of @xmath53 for a total snii rate of @xmath54 and a total gas mass of @xmath55 in the present galaxy ( qian & wasserburg 2001a ) .
this reference mass is also the total amount of ism typically swept by an snii remnant ( e.g. , thornton et al .
we take this as the standard reference mass of hydrogen to mix with the nucleosynthetic products of an snii @xmath3 or @xmath4 event .
we will use equation ( [ eps ] ) with fixed parameters @xmath56 , @xmath57 , and @xmath58 to discuss abundances in ump and mp stars .
all quantities in this equation will be calculated for the above reference mass of hydrogen .
the absolute yield of element e in units of mass for an @xmath3 or @xmath4 event , @xmath59 or @xmath60 , can be estimated from the relationships : @xmath61 where @xmath62 or @xmath63 is the average atomic mass for the e isotopes produced in an @xmath3 or @xmath4 event . in discussing the stellar observations , we consider that the observed stars formed from the immediate precursor ism with the homogenized @xmath0-abundances of that local region . these @xmath0-abundances are not considered to have been altered by the subsequent evolution of the star .
if the surface of a star is contaminated by some form of mass transfer from a companion ( or nearby ) snii , then the observed abundance pattern should represent the products of that snii ( an @xmath3 or @xmath4 event ) , but the abundances would be greatly enhanced from those with which the star formed .
for example , the extremely high ba abundances in cs 31082 - 001 ( crosses linked by a line ) and cs 22892 - 052 ( open circle ) shown in figure [ ba ] may have resulted from surface contamination by the snii explosion ( of the @xmath3 type ) of their respective binary companions ( qw ) .
the issue of what level of @xmath0-process enrichment is so large that an snii companion ( or neighbor ) appears to be required has been discussed by qw and will be considered below ( see [ syzbump ] ) . given the @xmath22-inventory , @xmath56 , and the yields of @xmath3 and @xmath4 events , @xmath57 and @xmath58 , the abundance of element e in a star , @xmath64 , can be calculated from equation ( [ eps ] ) .
the numbers of contributing @xmath3 and @xmath4 events for the star , @xmath45 and @xmath46 , are required for this calculation .
if a certain element is exclusively produced by @xmath3 or @xmath4 events , then @xmath45 or @xmath46 can be simply obtained from the observed abundance of this element in the star by using the yield of an @xmath3 or @xmath4 event for this element .
data on ba ( @xmath65 ) at @xmath66<-3 $ ] ( mcwilliam et al . 1995 ; mcwilliam 1998 ) indicate that the @xmath22-inventory of heavy @xmath0-elements is negligible ( wq ; see figure [ ba ] ) .
observations also show that the abundances of ba and above in ump stars very closely follow the solar @xmath0-abundance pattern ( sneden et al .
1996 , 2000 ; westin et al .
we consider that @xmath3 events are the exclusive source for the heavy @xmath0-elements above ba ( possibly including ba ) with a yield pattern identical to the corresponding part of the solar @xmath0-abundance pattern ( see [ phcomp ] and [ heavy ] ) .
then the @xmath3-yield of any heavy @xmath0-element above ba can be used to calculate the value @xmath45 for a star .
we use eu ( @xmath67 ) for this purpose as it is typically observed in ump and mp stars and its solar inventory is dominated by @xmath0-process contributions .
specifically , the solar @xmath0-abundance of eu , @xmath68 ( arlandini et al . 1999 ) , was contributed by @xmath69 @xmath3 events for a frequency of @xmath70 over a period of @xmath71 yr prior to solar system formation ( ssf ; see [ nhlsun ] ) .
thus @xmath72 . the value @xmath45 for a star
is then : @xmath73 as little or no fe is produced in @xmath3 events ( see [ model1 ] ) , we consider that the fe abundance at @xmath74<-1 $ ] results from contributions of @xmath4 events being added to the @xmath22-inventory : @xmath75 as snia were the dominant fe source at [ fe / h ] @xmath76 , only a fraction @xmath77 of the solar fe inventory @xmath78 ( anders & grevesse 1989 ) was contributed by snii @xmath4 events . for @xmath79 ( e.g. , timmes et al . 1995 ; qian & wasserburg 2001a ) and with @xmath80 contributing @xmath4 events for a frequency of @xmath81 over a period of @xmath71 yr prior to ssf ( see [ nhlsun ] ) , @xmath82 corresponding to [ fe / h]@xmath83 .
we take @xmath84 corresponding to [ fe / h]@xmath85 .
the @xmath22-inventory of fe is overwhelmed by fe addition from a few @xmath4 events .
thus the possible range in [ fe / h]@xmath42 has little effect on the determination of @xmath46 for mp stars . by using the above @xmath22-inventory and @xmath4-yield of fe
, the value @xmath46 for a star can be obtained from equation ( [ epsfe ] ) or from its equivalent in the [ fe / h ] notation : @xmath86}=10^{{\rm [ fe / h]}_p}+ n_l\times 10^{{\rm [ fe / h]}_l}.\ ] ] in all our works , snii @xmath3 and @xmath4 events are considered to occur stochastically with frequencies of @xmath87 and @xmath88 in some standard reference mass of hydrogen throughout the galaxy over galactic history ( wbg ; wq ; qian & wasserburg 2000 , 2001a , qw ) . at any given time @xmath89 since the onset of snii , the probability @xmath90 for a standard parcel of the ism to have had a number @xmath45 of @xmath3 events is given by the poisson distribution : @xmath91 where @xmath92 is the average number of occurrences for @xmath3 events at time @xmath89 .
for @xmath4 events a similar expression applies : @xmath93 the variances of the distributions in equations ( [ pnht ] ) and ( [ pnlt ] ) are @xmath94 and @xmath95 , respectively .
the square root of the variance , @xmath96 or @xmath97 , can be used as a measure of the variation in @xmath45 or @xmath46 at time @xmath89 : @xmath98 or @xmath99 . at @xmath100 yr ,
@xmath101 corresponding to @xmath102 to @xmath103 , while @xmath104 corresponding to [ fe / h ] @xmath105 to @xmath106 . for shorter times
, the dispersion in @xmath46 becomes much more prominent .
for example , at @xmath107 yr , @xmath108 corresponding to @xmath109 to @xmath110 , while @xmath111 corresponding to [ fe / h ] @xmath112 to @xmath113 .
the abundances of heavy @xmath0-elements ( e.g. , eu ) are a better measure of time because of the higher frequency of @xmath3 events , and hence much lower dispersion in @xmath45 .
thus it follows that there is a substantial dispersion in [ fe / h ] for a rather fixed abundance of any heavy @xmath0-element at any given time in the range @xmath114 yr .
conversely , a given [ fe / h ] value may correspond to a significant range of possible times .
we consider that this effect causes the scatter in abundances of heavy @xmath0-elements at @xmath11 } < -1.0 $ ] with the dispersion rapidly increasing at lower [ fe / h ] .
the dispersion and mean trends for abundances of heavy and light @xmath0-elements in ump and mp stars have been discussed in some detail by qian ( 2001 ) based on the same approach . as discussed in 1.2 , the complex history of mixing and transport of elements in the ism from which a star formed is represented by @xmath45 and @xmath46 for the star .
it is shown above that these numbers can be calculated from the observed eu and fe abundances in the star by using the @xmath3-yield of eu and the @xmath4-yield of fe . substituting the expressions for @xmath45 ( eq . [ [ eu ] ] ) and @xmath46 ( eq . [ [ epsfe ] ] or [ [ fe ] ] ) into the expression for the abundance of element e ( eq . [ [ eh ] ] or [ [ eps ] ] )
, we obtain @xmath115.\ ] ] as the yield of element e relative to eu for @xmath3 events , ( e / eu)@xmath43 , or to fe for @xmath4 events , ( e / fe)@xmath44 , is assumed to be fixed , the observed eu or fe abundance in the star contains all the prior history of mixing and transport for the part of element e coproduced with eu in @xmath3 events or with fe in @xmath4 events .
thus the fundamental problem in our approach to understand abundances in ump and mp stars is to establish the @xmath22-inventory composition and the yield templates of @xmath3 and @xmath4 events that underlie the observations .
these results will be presented in the form of @xmath56 , @xmath57 , and @xmath58 corresponding to @xmath116 ( [ fe / h ] @xmath9 ) , @xmath117 , and @xmath118 ( [ fe / h]@xmath119 ) .
we assume here that the solar abundances are representative of an average parcel of the ism at the time of ssf .
we use the values @xmath122 and @xmath123 to account for the part of the solar inventory that was contributed by snii over a total uniform production time of @xmath124 yr prior to ssf .
these values are directly related to the frequencies of @xmath3 and @xmath4 events that were derived from consideration of meteoritic data on the abundances of @xmath14hf ( with a lifetime @xmath125 yr ) and @xmath15i ( @xmath126 yr ) in the early solar system ( wbg ) .
the meteoritic data require that replenishment of fresh @xmath14hf in an average parcel of the ism must occur on a timescale commensurate with its lifetime , which corresponds to a frequency of @xmath127 for @xmath3 events in a standard reference mass of hydrogen . as discussed in
[ model2 ] , this frequency can be explained by considering the galactic snii rate per unit mass of gas and the total amount of ism typically swept up by an snii remnant .
although there may be a range of possible values for @xmath128 , the observed abundances of heavy @xmath0-elements in ump and mp stars appear to be well represented by the @xmath3-yields of these elements calculated by using @xmath69 .
for example , the lowest eu abundances observed in such stars ( e.g. , mcwilliam et al .
1995 ; burris et al .
2000 ) are remarkably close to the eu yield of a single @xmath3 event [ @xmath129 calculated from the solar @xmath0-inventory of eu for @xmath69 . in any case , a change in @xmath130 causes only a uniform shift in the @xmath3-yields of the relevant elements as the yield template is assumed here to be fixed .
such a shift will not affect the basic results presented here ( see [ model2][nhl ] ) .
the timescale for replenishment of fresh @xmath15i in an average parcel of the ism required by the meteoritic data is much longer than its lifetime and corresponds to a frequency of @xmath131 for @xmath4 events .
a decrease in the replenishment timescale ( corresponding to an increase in @xmath88 ) would cause an exponential increase in the abundance of @xmath15i in the early solar system ( ess ) .
many meteoritic measurements give a rather precise and very low value for this abundance [ @xmath132i/@xmath133i)@xmath134 ; e.g. , brazzle et al .
1999 ] , severely limiting any possible increase in @xmath88 .
thus we focus on the effect of decreasing @xmath135 from @xmath136 .
for @xmath80 the @xmath4-yield of fe is [ fe / h]@xmath119 .
a decrease in @xmath137 by a factor of 2 gives [ fe / h]@xmath138 , which means that there would be a lack of stars between [ fe / h ] @xmath9 and @xmath139 .
this is incompatible with the observations that show a continuous population at [ fe / h ] @xmath140 . in any case
, a change in @xmath137 again causes only a uniform shift in the @xmath4-yields of the relevant elements .
thus changes in the values of @xmath120 and @xmath121 will not affect the basic conclusions of this work .
but if the changes are large , they will violate the constraints indicated above .
the standard values @xmath122 and @xmath123 will be used to derive the results in this work . in this study
we are mainly concerned with the @xmath0-process . as discussed in [ model1 ] , there must be at least two distinct types of @xmath0-process events .
we associate both of these with snii , although they have not yet been proven to be the @xmath0-process sites .
it was argued that the other candidate sites , such as neutron star mergers , would have difficulty explaining the observed @xmath0-abundances in ump and mp stars if they were the principal source for @xmath0-nuclei ( qian 2000 ) .
extensive and very sophisticated parametric studies of high neutron density environments have been carried out to gain insights into the conditions that are required for the @xmath0-process ( kratz et al .
1993 ; hoffman , woosley , & qian 1997 ; meyer & brown 1997 ; freiburghaus et al .
these studies are usually directed towards approximating the overall solar @xmath0-abundance pattern by the @xmath0-process production in a prototypical environment .
however , there is as yet no sufficient guidance from these studies to explain the distinct types of @xmath0-process events discussed in [ model1 ] or to make reliable _ a priori _ predictions for the templates of @xmath0-process yields ( @xmath0-yields ) that can be used in galactic chemical evolution models . as also discussed in
[ model1 ] , the first generations of very massive stars formed from big bang debris produced a @xmath22-inventory of fe and associated elements but very little heavy @xmath0-nuclei .
the formation and evolution of these stars and their nucleosynthetic products have been a long - standing problem ( e.g. , ezer & cameron 1971 ) and recently received renewed attention ( e.g. , bromm , coppi , & larson 2000 ; heger et al .
2000 ; fryer et al . 2001 ; baraffe , heger & woosley 2001 ; bromm et al .
however , so far there is not sufficient information from theoretical models to predict the composition of the @xmath22-inventory assigned here to the first generations of stars .
our approach here is to determine some of the @xmath22-inventory composition and the @xmath0-yields of snii @xmath3 and @xmath4 events from the observed abundances in several selected stars based on the phenomenological model laid out in [ model1][nhl ] .
these results will be used in the model to calculate the abundances for ump and mp stars in general .
the calculated abundances will then be compared with observations .
one of the selected stars is the sun whose so - called cosmic " abundance pattern ( mostly derived from meteoritic studies ) has served as the key to understanding nucleosynthesis ( burbidge et al .
1957 ; cameron 1957 ) .
the part of the solar abundances required here is the @xmath0-process component ( @xmath0-component ) that is obtained by subtracting the @xmath10-process component ( @xmath10-component ) from the total solar inventory .
one of the better understood processes of stellar nucleosynthesis is the @xmath10-process ( see review by busso , gallino , & wasserburg 1999 ) .
the site of this process is well established observationally to be thermally pulsing agb stars .
theoretical models for the @xmath10-process are based on the well ( but not completely ) understood laws of stellar structure and evolution and the nuclear reactions in the stellar interior that produce the neutrons for slow capture by the pre - existing seed nuclei .
the nuclear physics involved both in neutron production and in @xmath10-processing has been the subject of extensive theoretical and experimental work .
many cross sections for important neutron capture reactions have been reliably measured .
the classical analysis of the @xmath10-process relies on mainly the nuclear physics input and adopts a suitable distribution of neutron exposures that in principle can be calculated from stellar models .
this approach is particularly successful for the so - called main @xmath10-component ( see kppeler , beer , & wisshak 1989 for a review ) .
the most recent @xmath10-process calculations of arlandini et al .
( 1999 ) have emphasized the dependences of @xmath10-process yields on the initial mass and metallicity of agb stars .
their calculations were embedded in very high quality stellar models with highly resolved structure of thermal pulses .
they showed that the classical main @xmath10-component was very well reproduced by averaging the yields for models of @xmath141 and @xmath142 agb stars with @xmath143 ( @xmath144 is the total mass fraction of all elements above he and @xmath145 is the solar value ) . however , there are still questions with regard to the one - dimensional stellar models used in the new @xmath10-process calculations , such as the conditions obtained at critical boundary layers in agb stars .
one fundamental problem is the formation of the @xmath146c pocket that is the dominant source of neutrons [ via @xmath147 $ ] in the new stellar models . by assuming that the mass of the @xmath146c pocket is constant for agb stars with different metallicities
, it was shown that the neutron - to - seed ratio for the @xmath10-process increases with decreasing metallicity , giving rise to preponderant production of heavier elements by low metallicity agb stars ( gallino et al .
this result is very plausible .
however , it appears to be in conflict with the recent observations of lp 625 - 44 ( aoki et al . 2000 ) .
this star has [ fe / h ] @xmath148 and is enormously enriched in heavy elements from ba and above .
the high enrichments in lp 625 - 44 were attributed to surface contamination by mass transfer from an evolved agb companion with the same metallicity [ fe / h ] @xmath148 ( aoki et al .
2000 ) and do not reflect significant @xmath10-process addition to the ism from which the star formed .
the abundance pattern from ba and above in this star matches that of the main @xmath10-component calculated by arlandini et al .
( 1999 ) from models of agb stars but with @xmath143 . as pointed out by aoki et al .
( 2000 ) , this abundance pattern is not in accord with models of low metallicity agb stars that have very high neutron - to - seed ratios : the observed pb / ba ratio is lower by two orders of magnitude than the value predicted by busso et al .
( 1999 ) from stellar models with [ fe / h ] @xmath148 . on the other hand ,
the observed abundances of sr , y , and zr relative to ba are a few orders of magnitude lower than what would be expected for stellar models with @xmath143 .
the @xmath10-process in low metallicity agb stars thus remains open to investigation , particularly with regard to the magnitude of the @xmath146c pocket .
we are fully aware that there have been remarkable advances in models of agb evolution and @xmath10-process nucleosynthesis over the past decade ( see busso et al .
1999 and references therein ) .
our purpose in the above discussion is not to decry the status of that field but to call attention to some uncertainties critical to our study .
the solar @xmath0-abundance of element e is calculated as @xmath149 in all cases where @xmath150 is small , the results for ( e / h)@xmath151 are sound and provide the basis for comparison of the observed @xmath0-abundance patterns in ump and mp stars with the solar @xmath0-component . in this work we are concerned with sr , y , zr , and ba for which @xmath152 with substantial uncertainties ( kppeler et al . 1989 ; arlandini et al .
for example , the value for ( sr / h)@xmath151 assigned by kppeler et al .
( 1989 ) has an uncertainty of 111% and is larger by factors of @xmath1533 than the values given by arlandini et al .
( 1999 ) . for y ,
the calculated @xmath10-process contribution exceeds its total solar inventory in the classical analysis by arlandini et al .
( 1999 ) and their stellar model calculations give a value for ( y / h)@xmath151 that is smaller by a factor of 3.5 than the value assigned by kppeler et al .
further , the division of the solar inventory into @xmath10-components and @xmath0-components is not well defined for sr , y , and zr as these elements can also be produced by the @xmath13-process that does not involve neutron capture but may occur in close connection with the @xmath0-process in snii ( woosley & hoffman 1992 ) . in the case of ba ,
kppeler et al . (
1989 ) gave only an upper limit for ( @xmath154ba / h@xmath155 and arlandini et al .
( 1999 ) have already found shifts in the main @xmath10-component of ba that result in a range of @xmath156 $ ] from 0.08 in the classical analysis to 0.19 in the stellar model calculations .
this clearly illustrates that care must be taken in using the solar @xmath0-abundances when the corresponding @xmath157 values are close to unity .
the @xmath157 values for the elements y and above that are used here correspond to the main @xmath10-component calculated by arlandini et al .
( 1999 ) from stellar models .
these results represent a significant step forward from the phenomenological approach of the classical analysis but they do not represent a full calculation of the integrated @xmath10-process production by agb stars over galactic history .
in addition to the main @xmath10-component , the classical analysis also introduces a so - called weak @xmath10-component that is important for sr and below ( e.g. , kppeler et al .
1989 ) . the @xmath157 value for sr used here is also taken from the stellar model calculations of arlandini et al .
( 1999 ) but includes the contribution from the weak @xmath10-component considered by these authors .
the @xmath157 values cited above and adopted for our initial calculations will be shown to result in a mismatch between the calculated sr , y , zr , and ba abundances and the observational data on mp stars .
it is widely recognized in the field that the abundances of sr , y and zr are problematic , particularly at low metallicities , relative to any current nucleosynthetic models ( m. busso & r. gallino , personal communication ) . while we also explore other explanations for the mismatch between our model results and data ( see [ dis ] ) , we feel that it is reasonable to consider the uncertainties in the solar @xmath10-component for sr , y , zr , and ba .
it is possible that the same uncertainties may also affect other elements whose @xmath157 values are not small .
the @xmath22-inventory and @xmath3 and @xmath4 yields are the key input to our model .
these are calculated in this section .
as addition of fe from a single @xmath4 event to the @xmath22-inventory would give [ fe / h ] @xmath158 ( see eq . [ [ fe ] ] ) , ump stars with [ fe / h ] @xmath9 could not have received any contributions from @xmath4 events ( i.e. , @xmath159 ) .
thus the abundances in these stars reflect only the addition of @xmath3-contributions to the @xmath22-inventory , providing a unique record for the @xmath22 and @xmath3 components .
equation ( [ eps ] ) can be simplified for ump stars as @xmath160 the value @xmath45 for a star can be calculated from the observed eu abundance in the star by using equation ( [ eu ] ) .
for example , eu data on the ump stars hd 122563 , hd 115444 , and cs 22892 - 052 ( westin et al .
2000 ; sneden et al .
2000 ) correspond to @xmath161 , 7 , and 36 ( see table [ syzph ] ) .
the abundances of element e other than eu in two ump stars with different @xmath45 provide a pair of equations in the form of equation ( [ ump ] ) with two unknown parameters @xmath56 and @xmath57 .
thus the data on two such ump stars ( singly and doubly primed quantities ) can be used to determine simultaneously the @xmath22-inventory and @xmath3-yield of element e : @xmath162 the results for sr , y , and zr obtained by using the data ( see table [ syzph ] ) on hd 115444 ( [ fe / h ] @xmath163 ) and cs 22892 - 052 ( [ fe / h ] @xmath164 ) are given in table [ phl ] .
the @xmath22-component of sr , y , and zr is associated with the prompt production of fe .
thus any difference in [ fe / h ] among ump stars should also be reflected by the @xmath22-component of these elements ( see below ) .
this indicates that the @xmath22-inventory of sr , y , and zr for hd 122563 with [ fe / h ] @xmath165 should be @xmath28 dex higher than the @xmath166 values for these elements calculated above by using the data on hd 115444 and cs 22892 - 052 , which have lower [ fe / h ] values ( by @xmath28 dex ) .
the observed eu abundance in hd 122563 gives @xmath161 for this star ( see table [ syzph ] ) .
the @xmath3-yields of sr , y , and zr calculated above are low , which suggests that the abundances of these elements in hd 122563 should essentially reflect their @xmath22-inventory for this star
. the observed abundances of sr , y , and zr in hd 122563 ( see table [ syzph ] ) are indeed @xmath1670.3 dex higher than the @xmath166 values in table [ phl ] . the level of metallicity necessary for the formation of the first generation of normal stars is not known _ a priori_. we have inferred from the observational data that normal astration first occurred at [ fe / h ] @xmath168 based on the phenomenological model .
recently , bromm et al . (
2001 ) have shown that a gas cloud with a baryon mass of @xmath169 and @xmath170 fails to undergo continued collapse and fragmentation whereas for @xmath171 the cloud can collapse and undergo vigorous fragmentation , leading to a large relative population of low mass clumps compared with the case of @xmath172 ( big bang debris ) .
the phenomenological and dynamical views thus appear to be in accord for the moment .
the difference in [ fe / h ] among ump stars referred to in our above discussion may have resulted from a range of values for the @xmath22-inventory of fe , possibly as wide as @xmath173}_p < -2.7 $ ] , over which normal astration may first have occurred .
we have chosen a nominal value of [ fe / h]@xmath85 in the above calculation of the @xmath22-component .
as we assign no eu or any other heavy @xmath0-element to the @xmath22-inventory , the possible range of [ fe / h]@xmath42 does not affect the calculation of @xmath45 nor that of @xmath174 for heavy @xmath0-elements produced by @xmath3 events .
however , elements such as sr , y , and zr associated with the @xmath22-inventory are directly related to the prompt fe production .
it follows that @xmath175 for these elements will be shifted with @xmath176 depending on the precise value of [ fe / h ] at which normal astration began . in comparing the @xmath22-components of element e for ump stars , it is therefore necessary to correct to a standard value of [ fe / h ] in order to obtain the @xmath22-inventory of element e relative to fe .
the @xmath175 values in table [ phl ] should be associated with [ fe / h]@xmath85 . if a ump star has [ fe / h ] @xmath177 , then its @xmath22-component has to be shifted by @xmath178 from the tabulated @xmath175 values as was done above for hd 122563 .
we have assigned the heavy @xmath0-elements above ba ( possibly including ba ) exclusively to @xmath3 events with a yield template that is identical to the corresponding part of the solar @xmath0-abundance pattern ( see [ heavy ] ) .
for these elements , including eu whose observed abundance is used to calculate @xmath45 for a star , @xmath179 and @xmath180 , where @xmath69 is the number of @xmath3 events that contributed to the solar inventory .
the @xmath3-yields of ba to au calculated this way have been given by qw and are repeated in table [ phl ] .
the @xmath3-yields of heavy @xmath0-elements other than eu may also be directly calculated from the observed abundances of these elements in any ump star without using their solar @xmath0-inventory .
for example , cs 22892 - 052 with @xmath181 ( see table [ syzph ] ) has @xmath182 ( sneden et al .
this gives @xmath183 in excellent agreement with the value @xmath184 obtained above for @xmath185 ( arlandini et al .
the long - lived chronometers @xmath186th and @xmath187u are produced along with the stable elements above ba ( possibly including ba ) in @xmath3 events . the @xmath3-yields of these two nuclei have been calculated by qw based on their solar @xmath0-abundances and are given in table [ phl ] .
of the stable elements above au , only pb has been observed in ump and mp stars with large uncertainties ( e.g. , sneden et al .
the @xmath3-yields of these elements are not given here .
besides sr , y , and zr , the light @xmath0-elements nb , ru , rh , pd , ag , and cd have been observed in the ump star cs 22892 - 052 , but with significantly lower abundances relative to the solar @xmath0-abundance pattern that is translated to pass through the eu data ( sneden et al .
this suggests that some light @xmath0-elements may be produced along with the heavy @xmath0-elements in @xmath3 events , but with relative yields of light to heavy @xmath0-elements significantly smaller than the corresponding solar @xmath0-abundance ratios .
if we consider that the @xmath22-inventory of nb , ru , rh , pd , ag , and cd is overwhelmed by contributions from the @xmath188 @xmath3 events assigned to cs 22892 - 052 ( see table [ syzph ] ) , then the @xmath3-yields of these elements can be estimated from their observed abundances , @xmath189 , as : @xmath190 .
similar results were obtained by qw who used the observed abundance of os instead of eu to calculate @xmath191 .
the @xmath22-inventory and @xmath3-yields derived above are summarized in table [ phl ] and shown in figure [ figphl ] . as discussed in
[ phcomp ] , the @xmath22-inventory and @xmath3-yields are calculated from data on two ump stars with different @xmath45 values . in this calculation , only the solar @xmath0-inventory of eu is needed to obtain the @xmath3-yield of eu . the yields for the other elements can be directly obtained from observations of ump stars although the yields of the other heavy @xmath0-elements can also be calculated by using their solar @xmath0-inventory .
in contrast , the @xmath4-yields will be calculated by using the solar @xmath0-inventory as an essential input .
this inventory was contributed by @xmath69 @xmath3 and @xmath80 @xmath4 events in addition to the @xmath22-inventory : @xmath192 the solar @xmath0-abundance @xmath193 given in table [ phl ] ( arlandini et al .
1999 ; see [ rsun ] ) is used in equation ( [ sun ] ) to calculate the @xmath4-yield of element e from the @xmath56 and @xmath57 values in table [ phl ] .
the resulting @xmath4-yields from this calculation are also given in table [ phl ] and shown in figure [ figphl ] .
note that @xmath194 for the heavy @xmath0-elements above ba ( possibly including ba ) as they have been exclusively assigned to @xmath3 events ( see [ phcomp ] ) .
the @xmath56 , @xmath57 , and @xmath58 values in table [ phl ] permit us to calculate the abundance of element e in any ump or mp star from the model described in [ intro ] by using only the observed eu and fe abundances to obtain @xmath45 and @xmath46 for the star ( see eqs . [ [ eps ] ] , [ [ eu ] ] , and [ [ fe ] ] ) .
the calculated abundances are compared with the observational data in this section .
the comparison of model results with observational data is the simplest for the heavy @xmath0-elements above ba ( possibly including ba ) as they have been exclusively assigned to @xmath3 events with a yield template identical to the corresponding part of the solar @xmath0-abundance pattern ( see [ phcomp ] ) . as a test of this hypothesis
, we consider the abundance ratios of other heavy @xmath0-elements ( e ) to eu .
these ratios should remain at the solar @xmath0-process values independent of [ fe / h ] ( i.e. , the occurrence of @xmath4 events ) if the heavy @xmath0-elements come only from @xmath3 events . based on the data of burris et al .
( 2000 ; open squares ) , johnson & bolte ( 2001 ; asterisks ) , and gratton & sneden ( 1994 ; filled triangles ) , the difference between the observed e / eu and the solar @xmath0-process value of arlandini et al .
( 1999 ) , [ e / eu]@xmath195 , is shown as a function of [ fe / h ] for ba , la , ce , pr , nd , sm , gd , and dy in figure [ heavyeu ] .
the data sets of burris et al .
( 2000 ) and johnson & bolte ( 2001 ) have 17 stars in common . figure [ heavyeu ] shows that : ( 1 ) ba / eu exhibits a regular increase with [ fe / h ] ; ( 2 ) la / eu appears to lie within @xmath28 dex of the solar @xmath0-process value over @xmath196\lesssim -1 $ ] ( with the exception of several data points ) ; ( 3 ) nd / eu shows a wide scatter with no identifiable trend ; ( 4 ) dy / eu appears to lie within @xmath28 dex of the solar @xmath0-process value over @xmath196\lesssim -1 $ ] ( with the exception of several data points ) ; ( 5 ) ce / eu , pr / eu , sm / eu , and gd / eu have only sparse data with no indication of any wide spread from the solar @xmath0-process values . in general , all those elements above ba ( except for nd ) appear to have ratios relative to eu that are roughly constant .
this demonstrates that they are to be associated exclusively with @xmath3 events , which is in accord with the model .
we can not explain the scatter in nd / eu .
further , we note that the observed values for la / eu , ce / eu , pr / eu , sm / eu , gd / eu , and dy / eu are close to the standard solar @xmath0-process values . of these elements ,
sm , eu , gd , and dy have predominant @xmath0-process contributions to their solar inventory @xmath197 ; see table [ phl ] ) .
however , la has @xmath198 from stellar model calculations and @xmath199 from the classical analysis ( arlandini et al .
these @xmath200 values correspond to a change in the solar @xmath0-component of la by a factor of 2.2 .
thus the present agreement between the observed la / eu and the corresponding solar @xmath0-process value from stellar model calculations of arlandini et al .
( 1999 ) is surprising , but encouraging . as la , ce , and pr have @xmath201 ( see table [ phl ] )
, it is clear that any significant @xmath10-process addition to the ism would drastically shift their abundances away from the nominal solar @xmath0-process values .
this is not observed .
we conclude that the observational data appear to justify two basic assumptions of our model : the @xmath3 events dominate the production of heavy @xmath0-elements from la and above and there is no significant @xmath10-process contribution to the ism prior to [ fe / h ] @xmath34 .
the new data also support and extend the earlier observational results that the @xmath0-abundance pattern above ba in ump stars is the same as that in the solar system ( sneden et al .
1996 , 2000 ; westin et al .
the need for more precise observational data on la and above is evident .
figure [ heavyeu ] shows that ba / eu exhibits systematic deviation from a constant value with increasing [ fe / h ] .
johnson & bolte ( 2001 ) have emphasized that the ba abundance data may reflect uncertainties in stellar atmosphere models .
these uncertainties may have caused the deviation for ba / eu as well as the wide scatter in nd / eu discussed above .
however , the extent to which artifacts in stellar atmosphere models would affect the abundance analysis for ba and other elements remains to be investigated . on the other hand ,
ba is special in that it is closer to the boundary between light and heavy @xmath0-elements .
this boundary has been considered to be at @xmath18 ( wbg ) .
it was also noted that any @xmath0-process model is expected to produce the abundance peak at @xmath18 with some overflow to higher masses ( qian et al .
1998 ) . in this case
, some contributions may be expected at and above ba from the @xmath4 events that are mainly responsible for the light @xmath0-elements .
a test of such overflow would be shifts in the abundance ratios of other heavy @xmath0-elements to eu with increasing @xmath4-contributions ( i.e. , increasing [ fe / h ] ) . as discussed above
, no such shifts have been observed for la ( see figure [ heavyeu ] ) .
thus the overflow in @xmath4 events must stop below la ( @xmath202 ) .
but ba is below la and may be liable to the influence of such overflow .
so the deviation of ba / eu from a constant value with increasing [ fe / h ] may be caused by the @xmath4-contributions to ba .
this possibility will be discussed further in [ syzbmp ] . the @xmath22-inventory and @xmath3-yields of sr , y , and zr in table [ phl ] are calculated from data on hd 115444 and cs 22892 - 052 ( both with [ fe / h ] @xmath9 ) .
these results are used to calculate the sr , y , and zr abundances in four other ump stars in the sample of burris et al .
for example , bd + 58@xmath2031218 has @xmath204 and [ fe / h ] @xmath205 corresponding to @xmath206 and @xmath207 . as its [ fe / h ] differs from [ fe / h]@xmath85 by @xmath208 dex , the corresponding @xmath22-components of sr , y , and zr must be increased from the @xmath166 values in table [ phl ] by the same amount ( see [ phcomp ] ) .
the calculated sr abundance for this star is @xmath209\approx 0.43 $ ] .
the values of @xmath210 and @xmath211 are obtained similarly .
these calculated abundances are within @xmath212 dex of the observational data @xmath213 , @xmath214 , and @xmath215 .
the difference between the model results and data , @xmath216 , is shown for the four ump stars in the sample of burris et al .
( 2000 ) in figure [ figsyzb ] ( filled squares ) . except for sr in hd 126587 , @xmath217 dex for all other cases .
we note that mcwilliam et al .
( 1995 ) gave @xmath218 for hd 126587 , which is 0.62 dex higher than the value of burris et al .
( 2000 ) . using the value of mcwilliam et al .
( 1995 ) would bring the calculated sr abundance of this star within 0.2 dex of the data .
with regard to ba in ump stars , we have calculated the abundances assuming pure @xmath3 contributions and using the observed @xmath219 to determine the value of @xmath45 .
the calculated values are in very good agreement with the observational data ( see filled squares in figure [ figsyzb ] ) .
there are two issues regarding the ba data at @xmath220 [ fe / h ] @xmath221 ( see figure [ ba ] ) .
the low ba abundances at @xmath220 [ fe / h ] @xmath222 compared with the @xmath3-yield of ba ( dotted line in figure [ ba ] ) were attributed to very small admixtures of @xmath3 contributions into clouds having almost pure @xmath22-inventory composition .
there is also a rapid rise of @xmath223 at @xmath224 [ fe / h ] @xmath221 . the dot - dashed line from the model ( see [ syzbmp ] ) that describes the trend of data at [ fe / h ] @xmath33 where both @xmath3 and @xmath4 events occur is extended to [ fe / h ] @xmath225 in figure [ ba ] .
the deviation ( by @xmath2262 dex ) of the data at @xmath224 [ fe / h ] @xmath221 from the tend at [ fe / h ] @xmath33 is evident .
this is the wall of rapid rise described by wq and was attributed to the onset of @xmath3 events with no fe production .
a problem exists as to how high this rise might be if it represents a sampling of the ism . as the @xmath3 yield of ba , @xmath227 ,
is well established and the frequency of @xmath3 events is @xmath16 , it is clear that for a ump star , @xmath223 should not represent more @xmath3 events than say @xmath228 corresponding to a period of @xmath229 yr during which an @xmath4 event should occur with very high probability . observed values of @xmath230 representing @xmath231 @xmath3 events but with no concomitant increase in [ fe / h ] , such as in the cases of cs 31082 - 001 ( crosses linked by a line in figure [ ba ] ) and cs 22892 - 052 ( open circle ) , are then attributed to contamination by an snii @xmath3 event in a binary and not to gross increases of @xmath0-abundances in the ism from which the stars formed ( qw ) .
surface contamination may also apply to some data at [ fe / h ] @xmath33 that lie far above the trend line in figure [ ba ]
. clearly , chemical enrichments by surface contamination in binary systems can not be ignored ( see preston & sneden 2001 ) .
here we focus on the discussion of mp stars .
the range of @xmath74<-1 $ ] for mp stars is considered to result from addition of fe from @xmath4 events to the @xmath22-inventory . in extending the discussion from ump stars with no contributions from @xmath4 events to mp stars with such contributions , estimates of the @xmath4-yields
play an essential role .
the @xmath4-yields , @xmath58 , are calculated from the values of @xmath193 , @xmath56 , and @xmath57 given in table [ phl ] .
the abundances of sr , y , and zr are then calculated from the model by using the @xmath22-inventory and @xmath3 and @xmath4 yields together with the observed @xmath219 and [ fe / h ] for the mp stars in the sample of burris et al .
for example , hd 122956 has @xmath232 and [ fe / h ] @xmath233 corresponding to @xmath234 and @xmath235 . the calculated sr abundance for this star is @xmath236\approx 0.63 $ ] .
the values of @xmath237 and @xmath238 are obtained similarly . by comparison ,
the observed values are @xmath239 , @xmath240 , and @xmath241 .
thus the calculated abundances ( especially for sr ) significantly differ from the data .
the difference between the model results and data , @xmath242 , is shown as a function of [ fe / h ] for the mp stars in the sample of burris et al .
( 2000 ) in figure [ figsyzb ] ( open squares ) .
it can be seen that the model systematically underestimates sr , y , and zr abundances at @xmath243 } < -1 $ ] . as these metallicities correspond to increasing contributions from @xmath4 events , we consider that this systematic deviation is possibly caused by errors in the @xmath4-yields of sr , y , and zr as determined from the solar @xmath0-inventory .
note that the solar @xmath0-abundance of sr assigned by arlandini et al .
( 1999 ) is saturated by the contributions from @xmath69 @xmath3 events and thus does not allow any @xmath4-contributions to sr . however , the solar @xmath0-abundances of y and zr assigned by the same authors allow substantial @xmath4-contributions to these two elements ( see table [ phl ] ) . as sr ,
y , and zr are close in atomic mass , it is rather difficult to explain why @xmath4 events can produce y and zr but no sr .
it seems more likely that the standard solar @xmath0-abundances of sr , y , and zr are too uncertain to give reliable @xmath4-yields for these elements . as discussed in
[ rsun ] , the solar @xmath0-process fraction @xmath244 $ ] for element e can have quite large errors when @xmath150 is close to unity as for sr , y , and zr ( see table [ phl ] ) .
we now consider that the solar @xmath0-inventory of sr , y , and zr given by arlandini et al .
( 1999 ) and adopted above may be increased by factors of 6.3 , 3.5 , and 3 , respectively , to obtain a `` corrected '' solar @xmath0-inventory @xmath245 and the corresponding @xmath246 values in table [ phl ] . with these @xmath246 values ,
the model results on sr , y , and zr abundances at @xmath247 } < -1 $ ] are generally in quite good agreement with data ( see filled circles in figure [ figsyzb ] ) .
the problems and issues that are raised by the hypothesized shifts in @xmath150 for sr , y , and zr will be discussed in [ dis ] .
it has been well known for some time that the abundances of sr , y , and zr in ump and mp stars can not be explained by any standard nucleosynthetic models ( m. busso & r. gallino , personal communication ) .
we note that the @xmath22-inventory , @xmath3 yields , and corrected @xmath4 yields of these elements given here follow approximately the same relative production pattern with a significant dip at y. this is not the case if the standard @xmath157 values for sr , y , and zr are used .
if we assign @xmath0-process production of ba exclusively to @xmath3 events , the ba abundance in an mp star can be calculated as @xmath248 , where @xmath249 can be obtained from the solar @xmath0-inventory of ba for @xmath69 or from the data on cs 22892 - 052 with essentially the same result ( see [ phcomp ] and table [ phl ] ) .
the difference between the calculated results and data of burris et al .
( 2000 ) , @xmath250 , is shown as a function of [ fe / h ] in figure [ figsyzb ] ( open squares ) .
note that @xmath251 lies significantly below zero at @xmath243 } < -1 $ ] .
this is the same as found for sr , y , and zr .
as indicated above , this result follows whether we calculate @xmath249 by using the data on a ump star or by attributing the standard solar @xmath0-inventory of ba from arlandini et al .
( 1999 ) to @xmath69 @xmath3 events .
if we consider that the solar @xmath0-inventory of ba assumed above is too small and that @xmath4 events contribute some ba , we may find a change that could eliminate the discrepancy .
using @xmath252 calculated from the data on cs 22892 - 052 ( see [ phcomp ] ) and increasing the solar @xmath0-inventory of ba by a factor of 2 from the value given by arlandini et al .
( 1999 ) , we obtain @xmath253 ( see eq . [ [ sun ] ] ) .
we then recalculate @xmath254 from equation ( [ eps ] ) by including the contributions to ba from @xmath4 events .
the recalculated @xmath251 is shown in figure [ figsyzb ] ( filled circles ) .
it can be seen that the model results with ba contributions from @xmath4 events represent data quite well over @xmath243 } < -1 $ ] .
this also accounts for the trend of ba / eu ( see [ heavy ] and figure [ heavyeu ] ) found by burris et al .
( 2000 ) and johnson & bolte ( 2001 ) : the shift in ba / eu with [ fe / h ] is caused by the increasing contributions to ba from @xmath4 events that do not produce eu .
the mean trend for evolution of ba abundance relative to fe at [ fe / h ] @xmath33 can be calculated from the values of @xmath249 , @xmath255 , and @xmath256 by assuming @xmath257 .
this is shown as the dot - dashed line in figure [ ba ] . as an independent test
, we may calculate @xmath258 from data on hd 126238 ( gratton & sneden 1994 ) that is not in the sample of burris et al .
( 2000 ) . in this calculation
the part of the solar fe inventory contributed by snii is used to determine the @xmath4-yield of fe , and hence @xmath46 , but the solar @xmath0-inventory of ba is not used .
the value of @xmath259 is calculated from the value of @xmath249 ( obtained from the data on cs 22892 - 052 ) and the observed value of @xmath223 in hd 126238 together with the values of @xmath45 and @xmath46 for this star .
the observed values of @xmath260 and [ fe / h ] @xmath261 for hd 126238 correspond to @xmath262 and @xmath263 . using @xmath264
, we obtain @xmath265 from the observed value of @xmath266 for hd 126238 .
this is in excellent agreement with the above value of @xmath267 .
it thus appears that the value of @xmath258 directly determined from data on a single mp star not in the sample of burris et al .
( 2000 ) without using the solar @xmath0-inventory of ba is in accord with the value obtained by changing this inventory to fit the data of burris et al .
( 2000 ) on a large sample of mp stars .
we consider the above direct calculation and the problem of fitting the ba data as some evidence indicating that the standard solar @xmath0-inventory of ba may require revision .
the required change in this inventory by a factor of 2 is not within the @xmath268 error band for @xmath156 $ ] given by arlandini et al .
( 1999 ) , but corresponds to a change of only 23% from their assigned value of @xmath269 to the corrected value of @xmath270 . to further test the model with corrected solar @xmath0-inventory of sr , y , zr , and ba
, we apply it to the data sets of magain ( 1989 ) and gratton & sneden ( 1994 ) that are distinct from the data set of burris et al .
the difference between the model results and data , @xmath271 , is shown for sr , y , zr , and ba as a function of [ fe / h ] in figure [ test ] .
it can be seen that by using the corrections made to the solar @xmath0-inventory for these elements based on the data set of burris et al .
( 2000 ) , the model describes the data sets of magain ( 1989 ) and gratton & sneden ( 1994 ) equally well ( @xmath217 dex in most cases ) .
thus we consider that the model presented above at least represents a very good parametric fit to the observational data on ump and mp stars from several groups .
it also seems possible that the model has deeper implications for @xmath10-process and @xmath0-process nucleosynthesis models and for the chemical evolution of the galaxy .
we have developed the consequences of a phenomenological model for the abundances of elements in the ism for [ fe / h ] @xmath272 .
this model assumes three components of nucleosynthetic contributions to @xmath0-process related elements .
one component is contributed by the first generations of very massive ( @xmath273 ) stars formed after the big bang .
these stars provide an initial or prompt inventory of fe and associated elements corresponding to [ fe / h]@xmath85 .
this @xmath22-inventory is distinctive and contains no heavy @xmath0-elements ( with @xmath6 ) . once the @xmath22-inventory is established , normal stellar populations with masses of @xmath274@xmath25 , which include the progenitors for two types of snii , begin to form at [ fe / h ] @xmath9 .
the two types of snii correspond to the other two components of the model .
they are the high frequency @xmath3 events that produce mainly heavy @xmath0-elements but no fe , and the low frequency @xmath4 events that produce mainly light @xmath0-elements ( with @xmath7 ) and fe .
the composition of the @xmath22-inventory and the yield templates of @xmath3 and @xmath4 events are assumed to be fixed and distinct from each other .
this allows the deconvolution of observational data to establish quantitatively the @xmath22-inventory and @xmath3 and @xmath4 yields .
the basic input for this procedure is the observational data on two ump stars and the solar @xmath0-abundances .
the element eu commonly observed in ump and mp stars is essentially a pure @xmath0-element .
as eu and all the other heavy @xmath0-elements above ba are attributed exclusively to @xmath3 events , it is convenient to use the observed eu abundance in a ump or mp star to identify the @xmath3 contributions to the abundances in the star ( any other heavy @xmath0-element above ba can also be used for this purpose ) .
the @xmath3-yield of eu is calculated from its solar @xmath0-abundance by using the number of contributing @xmath3 events for the solar inventory , which can be estimated from the frequency of these events .
the number of contributing @xmath3 events for a ump or mp star can then be obtained from the observed eu abundance in the star by using the @xmath3-yield of eu . for ump stars with [ fe / h ] @xmath275 [ fe / h]@xmath85
, their abundances only consist of @xmath22 and @xmath3 contributions . by using the data on two ump stars with [ fe / h ] @xmath9 but with different eu abundances ,
the @xmath22-inventory and @xmath3-yields of sr , y , and zr are simultaneously obtained .
the data on a third ump star give essentially the same results for the @xmath22-inventory when corrected to the same [ fe / h]@xmath42 value ( see [ phcomp ] ) .
these results are used in the model to calculate the sr , y , and zr abundances in other ump stars .
the calculated abundances are shown to be in very good agreement with the observational data ( see filled squares in figure [ figsyzb ] ) .
the @xmath22-inventory represents the sum of contributions from the first generations of very massive stars .
we consider that the @xmath22-inventory of sr , y , and zr reported here is reliable and may serve as a guide in evaluating models of nucleosynthesis for these stars .
the @xmath3-yields of heavy @xmath0-elements above ba other than eu can be calculated from the solar @xmath0-abundances as in the case of eu or from the observed abundances in a ump star by using the observed eu abundance to obtain the number of contributing @xmath3 events for the star .
these two approaches give effectively the same results as the @xmath0-abundance pattern above ba observed in ump stars is the same as that in the solar system .
for mp stars with @xmath276 < -1 $ ] , their fe abundances reflect contributions from @xmath4 events . in general , the abundances in these stars are governed by @xmath22 , @xmath3 , and @xmath4 contributions . for the heavy @xmath0-elements above
ba that only have @xmath3 contributions in our model , the abundance ratios of the other heavy @xmath0-elements to eu in mp stars are close to the solar @xmath0-process values assigned by arlandini et al .
( 1999 ) over a wide range in [ fe / h ] ( see figure [ heavyeu ] ) . for other elements that have @xmath4 contributions ,
their @xmath4-yields are estimated from their solar @xmath0-abundances , @xmath22-inventory , and @xmath3-yields by using the numbers of contributing @xmath3 and @xmath4 events for the solar inventory .
the abundances of these elements in an mp star can then be calculated from the three - component model by using the observed eu and fe abundances to obtain the numbers of contributing @xmath3 and @xmath4 events for the star . by comparing the model results with the observational data on @xmath27740 mp stars in the sample of burris et al .
( 2000 ) , it is found that to match the sr , y , zr , and ba data requires major changes in the solar @xmath0-abundances of these elements assigned by arlandini et al .
( 1999 ) ( see table [ phl ] and figure [ figsyzb ] ) . the model with a set of corrected solar @xmath0-abundances for these elements determined by fitting the data set of burris et al .
( 2000 ) is shown to describe the two different data sets of magain ( 1989 ) and gratton & sneden ( 1994 ) equally well over a wide range in [ fe / h ] ( see figure [ test ] ) .
all of the data sets on ba exhibit increases of ba / eu above the standard solar @xmath0-process value with increasing [ fe / h ] . this can be explained by the increasing contributions to ba from @xmath4 events that do not produce eu based on the model with the corrected solar @xmath0-abundance of ba . in summary , by using the @xmath22-inventory , @xmath3 yields , and @xmath4 yields ( see table [ phl ] and figure [ phl ] ) that are obtained from the data on two ump stars , the corrected solar @xmath0-abundances for sr , y , zr , and ba , and the standard solar @xmath0-abundances of arlandini et al .
( 1999 ) for other elements , it is possible to calculate the abundances of a large number of elements in a ump or mp star simply from the observed eu and fe abundances in the star over the range @xmath29}<-1 $ ] .
the fundamental question arises as to whether the corrected solar @xmath0-abundances for sr , y , zr , and ba can be justified in terms of nuclear astrophysical considerations .
all the other elements do not seem to require a significant shift in their solar @xmath0-inventory .
it is shown that the model is robust insofar as the relative yield patterns for @xmath3 and @xmath4 events are fixed .
changes in the absolute yields ( but not the relative yields ) do not change the patterns but only the numbers of contributing @xmath3 and @xmath4 events for a star .
the requirement for a large shift in the solar @xmath0-inventory for sr , y , zr , and ba then remains unchanged in this case . with regard to contributions from other sources ,
we note that the observational data on la , ce , and pr with solar @xmath10-process fractions of @xmath201 show no evidence of significant @xmath10-process contributions to the ism at [ fe / h ] @xmath272 .
thus possible @xmath10-process additions to the ism from agb stars are unimportant at [ fe / h ] @xmath272 and can not affect the required changes in the solar @xmath0-inventory for sr , y , zr , and ba . we note that extremely high @xmath10-process enrichment has been observed in lp 625 - 44 with [ fe / h ] @xmath148 ( aoki et al . 2000 ) .
however , this @xmath10-process enrichment does not reflect the composition of the ism at [ fe / h ] @xmath148 from which the star formed but is due to surface contamination by mass transfer from the binary companion of the star .
further , the observed [ fe / h ] in the star only indicates the formation time of the binary but does not correspond to the much later occurrence of @xmath10-processing in the agb companion .
it is possible that some other aspects of the model not considered in the preceding discussion cause the disagreement between the model results and observational data for sr , y , zr , and ba when the `` standard '' solar @xmath0-inventory for these elements is used .
for example , the relative yields of the pertinent elements may not be fixed as assumed but may vary from one @xmath3 or @xmath4 event to another . in this case , the @xmath3 and @xmath4 contributions to the abundances in a star can not be identified simply by using the observed abundances of eu and fe in the star .
the problem of abundances in ump and mp stars would then require a totally different approach .
unless confronted with new data that indicate otherwise , we adhere to the basic assumption of constant relative yields for @xmath3 or @xmath4 events and consider the following obvious alternatives to explain the disagreement referred to above : ( 1 ) there are some systematic shifts in the observational data provided by several groups for sr , y , zr , and ba as a function of [ fe / h ] , perhaps due to uncertainties in the stellar atmosphere models as suggested by johnson & bolte ( 2001 ) for ba ; or ( 2 ) there are intrinsic errors in the solar @xmath0-inventory for sr , y , zr , and ba .
the observed ba / eu shows a regular increase above the standard solar @xmath0-process value with increasing [ fe / h ] . however , the @xmath3-yield of ba calculated from the data on a ump star is in exact agreement with its solar @xmath0-abundance assigned by arlandini et al .
( 1999 ) if there are no @xmath4 contributions to ba .
this suggests that the observed shifts in ba / eu with increasing [ fe / h ] are the result of an artifact in abundance analysis that also depends on [ fe / h ] .
we find it difficult to understand how artifacts in analysis of the observational data would cause large shifts in abundance only for ba without having large effects on other elements .
if the ba problem is due to such an artifact , then the remaining inconsistency for sr , y , and zr may be accounted for by the acceptable @xmath10-process uncertainties , and hence errors in the solar @xmath0-inventory for these elements .
we emphasize that for sr , y , and zr , the @xmath157 values calculated from @xmath10-process models are all close to unity and lie in a region where serious uncertainties or errors may exist as has long been recognized by experts in @xmath10-process modeling .
if the ba problem is not due to artifacts in the abundance analysis , then the matter is not simply clarified without changing the standard solar @xmath0-abundance of ba .
the case of ba is similar to that of sr , y , and zr as @xmath278 is also close to unity .
however , ba lies in a region where the main @xmath10-component is dominant and is subject to more constraints than sr , y , and zr . at present
, the solution to this problem is not evident . however , there is a legitimate basis to consider that the solar @xmath0-component obtained by subtracting the @xmath10-component is subject to real uncertainties for elements such as sr , y , zr , and ba with predominant @xmath10-process contributions to their solar inventory .
the solar @xmath10-component used here corresponds to an average of the @xmath10-process yields from models of @xmath141 and @xmath142 agb stars with @xmath279 that best reproduces the classical main @xmath10-component ( arlandini et al . 1999 ) .
in fact , the solar @xmath10-component is the integral of long term galactic chemical evolution starting with low metallicity agb stars .
researchers in the field are fully aware of this matter .
the solar @xmath0-component derived from the subtraction procedure inherits all the uncertainties in the @xmath10-component calculated from models .
a reliable _ ab initio _ calculation of the solar @xmath0-component based on models of snii and galactic chemical evolution remains to be pursued .
critical attention to the investigation of @xmath10-process and @xmath0-process nucleosynthesis indicated above is required to sort out the problems . in any case , based on the observations of ump stars , the @xmath22-inventory produced by the first generations of very massive stars and yields of snii @xmath3 events appear to be well established .
this constitutes a prediction of what nucleosynthetic models for very massive zero - metallicity stars and for the common snii should produce . by using the three - component model with corrected solar @xmath0-abundances for sr , y , zr , and ba , and the standard values for other elements , the abundances of a large number of elements in stars with @xmath224 [ fe / h ] @xmath272 can be calculated very simply from the observed eu and fe abundances . whether this model is a reasonable physical description of galactic chemical evolution for @xmath0-elements or is a parametric description with alternative explanations
can not be answered at present .
we wish to thank the reviewers achim weiss and an unidentified person for their comments and for suggesting a more extensive treatment of the model .
in particular , achim weiss has raised several issues requiring discussion .
a new data set that appeared after the original submission and two additional data sets are now treated .
we have profited greatly from comments and continued enthusiastic support by roger blandford and the interest of marc kamionkowski .
maurizio busso has generously given aid and advice on the @xmath10-process problems considered here . a conversation with wal sargent stimulated our thinking about the mixing problem .
this work was supported in part by doe grants de - fg02 - 87er40328 and de - fg02 - 00er41149 ( y.z.q . ) and by nasa grant nag5 - 4083 ( g.j.w . ) , caltech division contribution 8741(1072 ) .
clepoch & + ( 1)&big bang + ( 2 ) & formation and explosion of very massive ( @xmath280 ) stars + & in regions of baryonic matter condensation with production of `` metals '' + & ( c , n , o , mg , al , si , ... , fe , ... , sr , y , zr ) + ( 3)&disruption of condensed regions by explosions of very massive stars , + & return of material with `` metals '' to the igm , and reionization of the igm + ( 4)&cooling and condensation of some of the baryonic matter in the igm + ( 5)&formation of normal stellar populations with masses of @xmath281@xmath25 + & when a metallicity of @xmath282}<-2.7 $ ] was reached + ( 6)&occurrence of the first snii @xmath3 events [ with a frequency of @xmath283 + & for a standard reference mass of hydrogen ] + ( 7)&coeval occurrence of snii @xmath3 and @xmath4 events [ with an @xmath4-frequency of + & @xmath284 for a standard reference mass of hydrogen ] + ( 8)&later occurrence of snia with major fe addition and evolution of low - mass agb + & stars resulting in substantial @xmath10-process contributions to the ism at [ fe / h ] @xmath285 + ccrrrrrrrr 26&fe & 4.51&@xmath296&5.03 & & & & & + 38&sr&0.13&@xmath297&@xmath296 & 1.64&0.95&0.35&2.44&0.69 + 39&y&@xmath298&@xmath299&@xmath106&1.12&0.92&@xmath300&1.67&0.72 + 40&zr&@xmath301&@xmath302&@xmath303&1.85&0.83&0.26&2.32&0.49 + 41&nb&@xmath296&@xmath304&@xmath305&0.56&0.85 & & & + 44&ru&@xmath296&@xmath306&@xmath307&1.60&0.32 & & & + 45&rh&@xmath296&@xmath308&@xmath309&1.03&0.14 & & & + 46&pd&@xmath296&@xmath310&@xmath311&1.42&0.46 & & & + 47&ag&@xmath296&@xmath308&@xmath312&1.14&0.20 & & & + 48&cd&@xmath296&@xmath313&@xmath314&1.42&0.52 & & & + 56&ba&@xmath296&@xmath315 & @xmath296&1.48&0.81&@xmath316&1.79&0.62 + 57&la&@xmath296&@xmath317&@xmath296&0.78&0.62 & & & + 58&ce&@xmath296&@xmath318&@xmath296&0.98&0.77 & & & + 59&pr&@xmath296&@xmath319&@xmath296&0.49&0.49 & & & + 60&nd&@xmath296&@xmath320&@xmath296&1.10&0.56 & & & + 62&sm&@xmath296&@xmath321&@xmath296&0.80&0.29 & & & + 63&eu&@xmath296&@xmath322&@xmath296&0.52&0.06 & & & + 64&gd&@xmath296&@xmath323&@xmath296&1.00&0.15 & & & + 65&tb&@xmath296&@xmath324&@xmath296&0.30&0.07 & & & + 66&dy&@xmath296&@xmath325&@xmath296&1.08&0.15 & & & + 67&ho&@xmath296&@xmath326&@xmath296&0.47&0.08 & & & + 68&er&@xmath296&@xmath327&@xmath296&0.87&0.17 & & & + 69&tm&@xmath296&@xmath328&@xmath296&0.07&0.13 & & & + 70&yb&@xmath296&@xmath317&@xmath296&0.78&0.33 & & & + 71&lu&@xmath296&@xmath329&@xmath296&0.02&0.20 & & & + 72&hf&@xmath296&@xmath330&@xmath296&0.39&0.56 & & & + 73&ta&@xmath296&@xmath331&@xmath296&@xmath332&0.41 & & & + 74&w&@xmath296&@xmath333&@xmath296&0.32&0.56 & & & + 75&re&@xmath296&@xmath334&@xmath296&0.25&0.09 & & & + 76&os&@xmath296&@xmath335&@xmath296&1.34&0.09 & & & + 77&ir&@xmath296&@xmath290&@xmath296&1.37&0.01 & & & + 78&pt&@xmath296&@xmath336&@xmath296&1.66&0.05 & & & + 79&au&@xmath296&@xmath321&@xmath296&0.80&0.06 & & & + 90&@xmath186th&@xmath296&@xmath337&@xmath296&0.18&0 & & & + 92&@xmath187u&@xmath296&@xmath338&@xmath296&@xmath339&0 & & & +
|
a model is presented that seeks to quantitatively explain the stellar abundances of @xmath0-process elements and other elements associated with the @xmath0-process sites .
it is argued that the abundances of all these elements in stars with @xmath1}<-1 $ ] can be explained by the contributions of three sources .
the sources are : the first generations of very massive @xmath2 stars that are formed from big bang debris and are distinct from snii , and two types of snii , the @xmath3 and @xmath4 events , which can occur only at [ fe / h ] @xmath5 .
the @xmath3 events are of high frequency and produce dominantly heavy ( @xmath6 ) @xmath0-elements but no fe ( presumably leaving behind black holes ) .
the @xmath4 events are of low frequency and produce fe and dominantly light ( @xmath7 ) @xmath0-elements ( essentially none above ba ) . by using the observed abundances in two ultra - metal - poor stars and the solar @xmath0-abundances , the initial or prompt inventory of elements produced by the first generations of very massive stars and the yields of @xmath3 and @xmath4 events can be determined .
the abundances of a large number of elements in a star can then be calculated from the model by using only the observed eu and fe abundances . to match the model results and the observational data for stars with
@xmath8}<-1 $ ] requires that the solar @xmath0-abundances for sr , y , zr , and ba must be significantly increased from the standard values .
no such changes appear to be required for all other elements .
if the changes in the solar @xmath0-abundances for sr , y , zr , and ba are not permitted , the model fails at @xmath8}<-1 $ ] but still works at [ fe / h ] @xmath9 for these four elements . by using the corrected solar @xmath0-abundances for these elements
, good agreement is obtained between the model results and data over the range @xmath8}<-1 $ ] .
no evidence of @xmath10-process contributions is found in this region , but all the observational data in this region now show regular increases of ba / eu above the standard solar @xmath0-process value . whether the solar @xmath0-components of sr , y , zr , and ba used here to obtain a fit to the stellar data can be reconciled with those obtained from solar abundances by subtracting the @xmath10-components calculated from models
is not clear .
|
measurements of galaxy rotation curves form the foundation of much of our understanding of galaxy formation , structure , and dynamics ( e.g. , binney & merrifield 1998 ; binney & tremaine 2008 ; mo , van den bosch , & white 2010 ) . the current knowledge of galaxy rotation is based entirely on observations of doppler shifts in radiation from galaxies .
this yields only one coordinate of motion , the los velocity .
if a galaxy rotates , and is not viewed edge - on , it will also rotate in the plane of the sky . until now
, the implied pms have generally been undetectable , given the available observational capabilities .
however , the observational capabilities have steadily advanced .
we present here new results for the lmc that constitute the first detailed measurement and analysis of the large - scale rotation field of any galaxy in all three dimensions .
the hubble space telescope ( hst ) provides a unique combination of high spatial resolution , long - term stability , exquisite instrument calibrations , and ever - increasing time baselines . over the past decade , this has opened up the local group of galaxies to detailed pm studies .
these studies have focused primarily on the satellites of the milky way ( kallivayalil et al .
2006a , hereafter k06 ; kallivayalil , van der marel & alcock 2006b ; piatek & prior 2008 , and references therein ; pryor , piatek & olszewski 2010 ; lpine et al . 2011 ; sohn et al .
2013 ; boylan - kolchin et al .
more recently it has even become possible to go out as far as m31 ( sohn et al .
2012 ; van der marel et al .
2012a , b ) .
all of these studies have aimed at measuring the systemic center - of - mass ( com ) motion of the target galaxies , and not their internal kinematics . so typically , only 13 different fields were observed in any given galaxy . by contrast , a study of internal kinematics requires , in addition to high pm accuracy , a larger number of different fields spread out over the face of the galaxy .
in k06 we presented a detailed pm study of the lmc .
we used hst to observe 21 fields centered on background quasars , in two epochs separated by a median baseline of 1.9 years .
the distribution of observed fields extends to 4@xmath4 from the lmc center ( @xmath5 for an assumed distance of @xmath6 , i.e. , @xmath7 ; freedman et al .
2001 ) . from the data we derived the average pm of the stars in each field .
we used this to estimate the pm of the lmc com . in besla
et al . ( 2007 ) our team studied the implied orbit of the magellanic clouds , and argued that they may be falling into the milky way for the first time .
the data also allowed us to detect the pm rotation of the lmc at @xmath8 significance .
the rotation sense and magnitude were found to be consistent with the detailed predictions for the lmc pm rotation field presented by van der marel et al .
( 2002 ; hereafter vdm02 ) , based on the observed los rotation field of carbon stars . piatek et al .
( 2008a , hereafter p08 ) performed a more sophisticated reanalysis of our k06 data , including small corrections for charge - transfer efficiency ( cte ) losses .
this yielded better pm consistency between fields , but implied a similar pm for the lmc com .
p08 used their measurements to derive the first crude pm rotation curve for the lmc , assuming fixed values for the dynamical center and disk orientation .
however , their inferred rotation amplitude @xmath9 appears unphysically high , exceeding the known rotation of cold hi gas ( kim et al .
1998 ; olsen & massey 2007 ) by @xmath10 .
so better data are needed to accurately address the pm rotation of the lmc .
we recently presented a third epoch of hst pm data for 10 fields ( kallivayalil et al .
2013 ; hereafter paper
i ) , increasing the median time baseline to 7.1 years . for these fields we obtained a median per - coordinate pm uncertainty of only 7 km / s ( 0.03 mas / yr ) , which is a factor 34 better than in k06 and p08 .
this corresponds to @xmath11% of the lmc rotation amplitude .
as we show in the present paper , these data are sufficient to map out the lmc pm rotation field in detail , yielding new determinations of the lmc dynamical center , disk orientation , and rotation curve .
the lmc is a particularly interesting galaxy for which to perform such a study . at a distance of only @xmath12 ,
it is one of nearest and best - studied galaxies next to our own milky way ( e.g. , westerlund 1997 ; van den bergh 2000 ) .
it is a benchmark for studies on various topics , including stellar populations and the interstellar medium , microlensing by dark objects , and the cosmological distance scale . as nearby companion of the milky way , with significant signs of interaction with the small magellanic cloud ( smc ) ,
the lmc is also an example of ongoing hierarchical structure formation .
for all these applications it is important to have a solid understanding of the lmc structure and kinematics .
the current state of knowledge about the kinematics of the lmc was reviewed recently by van der marel , kallivayalil & besla ( 2009 ) .
studies of the los velocities of many different tracers have contributed to this knowledge .
the kinematics of gas in the lmc has been studied primarily using hi ( e.g. , kim et al .
1998 ; olsen & massey 2007 ; olsen et al .
2011 , hereafter o11 ) .
discrete lmc tracers which have been studied kinematically include star clusters ( e.g. , schommer et al .
1992 ; grocholski et al .
2006 ) , planetary nebulae ( meatheringham et al .
1988 ) , hii regions ( feitzinger , schmidt - kaler & isserstedt 1977 ) , red supergiants ( prevot et al .
1985 ; massey & olsen 2003 ; o11 ) , red giant branch ( rgb ) stars ( zhao et al . 2003 ; cole et al .
2005 ; carrera et al .
2011 ) , carbon stars and other asymptotic giant branch ( agb ) stars ( e.g. , kunkel et al . 1997 ; hardy et al . 2001
; vdm02 ; olsen & massey 2007 ; o11 ) , and rr lyrae stars ( minniti et al .
2003 ; borissova et al .
2006 ) . for the majority of tracers ,
the line - of - sight velocity dispersion is at least a factor @xmath13 smaller than their rotation velocity .
this implies that on the whole the lmc is a ( kinematically cold ) disk system .
specific questions that can be addressed in a new way through a study of the lmc pm rotation field include the following : * what is the stellar dynamical center of the lmc , and does this coincide with the hi dynamical center ?
it has long been known that different measures of the lmc center ( e.g. , center of the bar , center of the outer isophotes , hi dynamical center , etc . )
are not spatially coincident ( e.g. , van der marel 2001 , hereafter vdm01 ; cole et al . 2005 ) , but a solid understanding of this remains lacking .
* what is the orientation under which we view the lmc disk ?
past determinations of the inclination angle and the line - of - nodes position angle have spanned a significant range , and the results from different studies are often not consistent within the stated uncertainties ( e.g. , van der marel et al .
knowledge of the orientation angles is necessary to determine the face - on properties of the lmc , with past work indicating that the lmc is not circular in its disk plane ( vdm01 ) .
* what is the pm of the lmc com , which is important for understanding the lmc orbit with respect to the milky way ?
we showed in paper i that the observational pm errors are now small enough that they are not the dominant uncertainty anymore .
instead , uncertainties in our knowledge of the geometry and kinematics of the lmc disk are now the main limiting factor . *
what is the rotation curve amplitude of the lmc ?
previous studies that used different tracers or methods sometimes obtained inconsistent values ( e.g. , p08 ; o11 ) .
the rotation curve amplitude is directly tied to the mass profile of the lmc , which is an important quantity for our understanding of the past orbital history of the lmc with respect to the milky way ( paper i ) . * what is the distance of the lmc ?
uncertainties in this distance form a key limitation in our understanding of the hubble constant ( e.g. , freedman et al .
comparison of the pm rotation amplitude ( in mas / yr ) and the los rotation amplitude ( in km / s ) can in principle yield a kinematical determination of the lmc distance that bypasses the stellar evolutionary uncertainties inherent to other methods ( gould 2000 ; van der marel et al .
2009 ) . in paper i of this series we presented our new third epoch observations ,
and we analyzed all the available hst pm data for the lmc ( and the smc ) .
we included a reanalysis of the earlier k06/p08 data , with appropriate corrections for cte losses .
we used the data to infer an improved value for the pm and the galactocentric velocity of the lmc com , and we discussed the implications for the orbit of the magellanic clouds with respect to the milky way ( and in particular whether or not the clouds are on their first infall ) . in the present paper
we use the pm data from paper i to study the internal kinematics of the lmc .
the outline of this paper is as follows .
section [ s : model ] discusses the pm rotation field , including both the data and our best - fit model .
section [ s : los ] presents a new analysis of the los kinematics of lmc tracers available from the literature . by including the new constraints from the pm data , this analysis yields a full three - dimensional view of the rotation of the lmc disk .
section [ s : param ] discusses implications of the results for our understanding of the geometry , kinematics , and structure of the lmc .
this includes discussions of the galaxy distance and systemic motion , the dynamical center and rotation curve , the disk orientation and limits on precession and nutation , and the galaxy mass .
we also discuss how the rotation of the lmc compares to that of other galaxies .
section [ s : conc ] summarizes the main conclusions .
we use the pm data presented in table 1 of paper i as the basis of our study .
the data consist of positions @xmath14 for 22 fields , with measured pms @xmath15 in the west and north directions , and corresponding pm uncertainties @xmath16 .
there are 10 `` high - accuracy '' fields with long time baselines ( @xmath17 years ) and three - epochs of data , and 12 `` low - accuracy '' fields with short time baselines ( @xmath13 years ) and two - epochs of data .
the pm measurement for each field represents the average pm of @xmath18 lmc stars with respect to one known background quasar .
the number of well - measured lmc stars varies by field , but is in the range 8129 , which a median @xmath19 .
the field size for each pm measurement corresponds to the footprint of the hst acs / hrc camera , which is @xmath20 arcmin .
this is negligible compared to the size of the lmc itself , which extends to a radius of @xmath21@xmath22 ( vdm01 ; saha et al .
2010 ) .
figure [ f : obsvar ] illustrates the data , by showing the spatially variable component of the observed pm field , @xmath23 , where the constant vector @xmath24 mas / yr . this vector is the best - fit pm of the lmc com as derived later in the present paper , and as discussed in paper i. clockwise motion is clearly evident .
the goal of the subsequent analysis is to model this motion to derive relevant kinematical and geometrical parameters for the lmc .
= 0.8 to interpret the lmc pm observations one needs a model for the pm vector @xmath25 as a function of position on the sky .
the pm model can be expressed as a function of equatorial coordinates , @xmath26 , or as a function of polar coordinates , @xmath27 , where @xmath28 is the angular distance from the lmc com and @xmath29 is the corresponding position angle measured from north over east .
generally speaking , the model can be written as a sum of two vectors , @xmath30 , representing the contributions from the systemic motion of the lmc com and from the internal rotation of the lmc , respectively .
consider first the contribution from the systemic motion .
the three - dimensional velocity that determines how the lmc com moves through space is a fixed vector .
however , the projection of this vector onto the west and north directions depends on where one looks in the lmc .
this introduces an important spatial variation in the pm field , due to several different effects , including : ( i ) only a fraction @xmath31 of the lmc transverse velocity is seen in the pm direction ; ( ii ) a fraction @xmath32 of the lmc los velocity is also seen in the pm direction ; and ( iii ) the directions of west and north are not fixed in a zenithal projection centered on the lmc , due to the deviation of @xmath14 contours from an orthogonal grid near the south galactic pole ( see figure 4 of van der marel & cioni 2001 , hereafter vdmc01 ) . as a result , one can write @xmath33 . the first term is the constant pm of the lmc com , measured at the position of the com .
the second term is the spatially varying component of the systemic contribution , which can be referred to as the `` viewing perspective '' component . to describe the component of internal rotation , we assume that the lmc is a flat disk with circular streamlines .
this is the same approach that has been used successfully to model los velocities in the lmc ( e.g. , vdm02 ; o11 ) .
however , it should be kept in mind that this model is only approximately correct .
the lmc is not circular in its disk plane ( vdm01 ) ,
so the streamlines are not expected to be exactly circular .
fortunately , the gravitational potential is always rounder than the density distribution , so circular streamlines should give a reasonable low - order approximation .
also , the modest @xmath34 of the lmc indicates that its disk is not particularly thin ( vdm02 ) .
so the flat - disk model should be viewed as an approximation to the actual ( three - dimensional ) velocity field as projected onto the disk plane . at any point in the disk ,
the relation between the transverse velocity @xmath35 in km / s and the pm @xmath36 in mas / yr is given by @xmath37 , where @xmath38 is the distance in kpc .
the distance @xmath38 is not the same for all fields , and is not the same as the distance @xmath39 of the lmc com .
the lmc is an inclined disk , so one side of the lmc is closer to use than the other .
this has been quantified explicitly by comparing the relative brightness of stars on opposite sides of the lmc ( e.g. , vdmc01 ) .
the analytical expressions for the pm field thus obtained , @xmath40 were presented in vdm02 .
we refer the reader to that paper for the details of the spherical trigonometry and linear algebra involved .
the following model parameters uniquely define the model : * the projected position @xmath41 of the lmc com , which is also the dynamical center of the lmc s rotation . * the orientation of the lmc disk , as defined by the inclination @xmath42 ( with @xmath43 defined as face - on ) and the position angle @xmath44 of the line of nodes ( the intersection of the disk and sky planes ) , measured from north over east . *
the pm of the lmc com , @xmath45 , expressed in the heliocentric frame ( i.e. , not corrected for the reflex motion of the sun ) . *
the heliocentric los velocity of the lmc com , @xmath46 , expressed in angular units ( for which we use mas / yr throughout this paper ) . * the rotation curve in the disk ,
@xmath47 , expressed in angular units . here
@xmath48 is the radius in the disk in physical units , and @xmath49 .
( along the line of nodes , @xmath50 ; in general , the lmc distance must be specified to calculate the radius in the disk is in physical units ) .
the first two bullets define the geometrical properties of the lmc , and the last three bullets its kinematical properties .
figures 10a , b of vdm02 illustrate the predicted morphology of the pm fields @xmath51 and @xmath52 for a specific lmc model tailored to fit the los velocity field .
these two components have comparable amplitudes .
the spatially variable component of the observed pm field @xmath53 in figure [ f : obsvar ] provides an observational estimate of the sum @xmath54 ( compare eq .
[ [ pmfield ] ] ) .
the pm field is defined by the variation of two components of motion over the face of the lmc . by contrast , the los velocity field is defined by the variation of only one component of motion .
the pm field therefore contains more information , and has more power to discriminate the parameters of the model .
as we will show , important constraints can be obtained with only 22 pm measurements , ) , so the shot noise is a factor @xmath55 smaller . moreover ,
his simulation had a central velocity dispersion in the disk of @xmath56 , which is higher than the value @xmath57 typical for the population that dominates the mass ( vdm02 ) .
as a result , his assumed random measurement uncertainties were a factor @xmath58 too high . ] whereas los velocity studies require hundreds or thousands of stars .
the following simple arguments show that knowledge of the full pm field in principle allows unique determination of all model parameters , without degeneracy .
* the dynamical center @xmath41 is the position around which the spatially variable component of the pm field has a well - defined sense of rotation .
* the azimuthal variation of the pm rotation field determines both of the lmc disk orientation angles @xmath59 .
perpendicular to the line of nodes ( i.e. , @xmath60 ) , all of the rotational velocity @xmath61 in the disk is seen as a pm ( and none is seen along the los ) . by contrast , along the line of nodes ( i.e. , @xmath62 or @xmath63 ) , only approximately @xmath64 is seen as a pm ( and approximately @xmath65 is seen along the los ) .
the near and far side of the disk are distinguished by the fact that velocities on the near side imply larger pms . *
the pm of the lmc com , @xmath45 , is the pm at the dynamical center .
* the systemic los velocity @xmath66 in angular units follows from the radially directed component of the pm field .
a fraction @xmath67 is seen in this direction ( appearing as an `` inflow '' for @xmath68 and an `` outflow '' for @xmath69 ) .
this component is almost perpendicular to the more tangentially oriented component induced by rotation in the lmc disk , so the two are not degenerate .
however , the radially directed component is small near the galaxy center ( e.g. , @xmath70 for @xmath71 ) , so exquisite pm data would be required to constrain @xmath66 with meaningful accuracy . *
the rotation curve @xmath47 in angular units follows from the pms along the line - of - nodes position angle @xmath44 .
llrrr @xmath72 & deg & @xmath73 & @xmath74 & @xmath75 + @xmath76 & deg & @xmath77 & @xmath78 & @xmath79 + @xmath42 & deg & @xmath80 & @xmath81 & @xmath82 + @xmath44 & deg & @xmath83 & @xmath84 & @xmath85 + @xmath86 & mas / yr & @xmath87 & @xmath88 & @xmath89 + @xmath90 & mas / yr & @xmath91 & @xmath92 & @xmath93 + @xmath94 & km / s & @xmath95 & @xmath96 & @xmath97 + @xmath98 & mas / yr & @xmath99 & @xmath100 & @xmath101 + @xmath102 & km / s & @xmath103 & @xmath104 & @xmath105 + @xmath106 & km / s & @xmath107 & @xmath108 & @xmath109 + @xmath110 & km / s & @xmath107 & @xmath111 & @xmath112 + @xmath113 & & @xmath114 & @xmath115 & @xmath116 + @xmath39 & kpc & @xmath117 & @xmath117 & @xmath117 + = 0.8 by contrast , full knowledge of the los velocity field does _ not _ constrain all the model parameters uniquely .
specifically , there is strong degeneracy between three of the model parameters ( see vdm02 ) : the rotation curve @xmath61 , the inclination angle @xmath42 ( since the observed los velocity component is approximately @xmath65 ) , and the component @xmath118 of the transverse com velocity vector @xmath119 projected onto the line of nodes ( which adds a solid - body component to the observed rotation ) . so the rotation curve can only be determined from the los velocity field if @xmath42 and @xmath118 are assumed to be known independently .
typically ( e.g. , vdm02 ; o11 ) , @xmath42 has been estimated from geometric methods ( e.g. , vdmc01 ) and @xmath118 from proper motion studies ( e.g. , k06 ) .
it should be noted that the transverse com velocity component @xmath120 in the direction perpendicular to the line of nodes _ is _ determined uniquely by the los velocity field , as is the position angle @xmath44 of the line of nodes itself .
and of course , the systemic los velocity @xmath94 is determined much more accurately by the los velocity field than by the pm field .
an important difference between the two observationally accessible fields is that the pm field constrains velocities in angular units ( mas / yr ) , whereas the los velocity field constrains the same velocities in physical units ( km / s ) .
hence , comparison of the results for e.g. @xmath61 or @xmath120 from the two fields constrains the lmc distance @xmath39 .
this is discussed further in section [ ss : dist ] . in our earlier analysis of k06
, we treated @xmath45 as the only free parameters to be determined from the pm data .
all other quantities were kept fixed to estimates previously obtained either by vdm02 from a study of the lmc los velocity field , by vdmc01 from a study of the lmc orientation angles , or by freedman et al .
( 2001 ) from a study of the lmc distance .
p08 took the same approach , but as discussed in section [ s : intro ] , they did treat the rotation curve @xmath61 as a free function to be determined from the data . keeping model parameters fixed a priori is reasonable when only limited data is available .
however , this does have several undesirable consequences .
first , it does not use the full information content of the pm data , which actually constrains the parameters independently .
second , it opens the possibility that parameters are used that are not actually consistent with the pm data . and
third , it leads to underestimates of the errorbars on the lmc com pm @xmath45 , since the uncertainties in the geometry and rotation of the lmc are not propagated into the answers ( as discussed in paper i ) . the three - epoch pm data presented in paper i have much improved quality over the two - epoch measurements presented by k06a and p08 , as evident from figure [ f : obsvar ] .
we therefore now treat all of the key parameters that determine the geometry and kinematics of the lmc as free parameters to be determined from the data .
there are @xmath121 lmc fields , and hence @xmath122 observed quantities ( there are two pm coordinates per field ) . by comparison
, the model is defined by the 7 parameters @xmath123 and the one - dimensional function @xmath47 .
the rotation curves of galaxies follow well - defined patterns , and are therefore easily parameterized with a small number of parameters .
we use a very simple form with two parameters @xmath124 \label{rotparam}\ ] ] ( similar to p08 and o11 ) .
this corresponds to a rotation curve that rises linearly to velocity @xmath125 at radius @xmath126 , and stays flat beyond that .
the quantity @xmath127 is the rotation amplitude expressed in angular units . later in section [ ss : rotcurve ]
we also present unparameterized estimates of the rotation curve @xmath61 . sticking with the parameterized form for now , we have an overdetermined problem with more data points ( @xmath128 ) than model parameters ( @xmath129 ) , so this is a well - posed mathematical problem .
we also know from the discussion in section [ ss : pmvslos ] that the model parameters should be uniquely defined by the data without degeneracy .
so we proceed by numerical fitting of the model to the data . to fit the model we define a @xmath130 quantity @xmath131 ^ 2 + \nonumber \\ & & \left [ ( \mu_{n,{\rm obs},i } - \mu_{n,{\rm mod},i } ) / \delta \mu_{n,{\rm obs},i } \right ] ^2 \label{chisqpm}\end{aligned}\ ] ] that sums the squared residuals over all @xmath132 fields .
we minimize @xmath133 as function of the model parameters using a down - hill simplex routine ( press et al .
multiple iterations and checks were built in to ensure that a global minimum was found in the multi - dimensional parameter space , instead of a local minimum .
llrr @xmath72 & deg & @xmath134 & @xmath134 + @xmath76 & deg & @xmath135 & @xmath135 + @xmath42 & deg & @xmath136 & @xmath136 + @xmath44 & deg & @xmath137 & @xmath138 + @xmath86 & mas / yr & @xmath139 & @xmath140 + @xmath90 & mas / yr & @xmath141 & @xmath142 + @xmath94 & km / s & @xmath95 & @xmath143 + @xmath106 & km / s & @xmath144 & @xmath145 + @xmath110 & km / s & @xmath146 & @xmath147 + @xmath113 & & @xmath148 & @xmath149 + @xmath39 & kpc & @xmath117 & @xmath117 + once the best - fitting model parameters are identified , we calculate error bars on the model parameters using monte carlo simulations . many different pseudo data sets
are created that are analyzed similarly to the real data set .
the dispersions in the inferred model parameters are a measure of the @xmath150 random errors on the model parameters .
each pseudo data set is created by calculating for each observed field the best - fit model pm prediction , and by adding to this random gaussian deviates .
the deviates are drawn from the known observational error bars , multiplied by a factor @xmath151 . here
@xmath152 is the @xmath153 value of the best - fit model , and @xmath154 is the number of degrees of freedom , with @xmath155 the number of parameters ( if any ) that are not optimized in the fit . in practice
we find that @xmath152 is somewhat larger than @xmath156 , indicating that the actual scatter in the data is slightly larger than what is accounted for by random errors .
this is not surprising , given the complexity of the astrometric data analysis and the relative simplicity of the model .
the approach used to create the pseudo - data ensures that the actual scatter is propagated into the final uncertainties on the model parameters .
it is known from los velocity studies that @xmath157 ( vdm02 ) , and from stellar population studies that @xmath158 kpc ( @xmath159 ; freedman et al .
2001 , but to be conservative , we use the older freedman et al .
( 2001 ) distance estimate throughout this paper . ] ) .
so @xmath94 is known to @xmath160 accuracy and @xmath38 to @xmath161 accuracy .
not surprisingly , we have found that the pm data can not constrain the model parameter @xmath66 with similar accuracy .
therefore , we have kept @xmath66 fixed in our analysis to the value implied by existing knowledge . at @xmath162 , 1
mas / yr corresponds to @xmath163 .
hence , @xmath164 mas / yr . the uncertainty in this value was propagated into the analysis by using randomly drawn @xmath66 values in the fitting of the different monte - carlo generated pseudo - data sets .
table [ t : param ] lists the parameters of the best - fit model and their uncertainties .
these parameters are discussed in detail in section [ s : param ] .
figure [ f : obsrot ] shows the data - model comparison for the best fit . for this figure
, we subtracted the systemic velocity contribution @xmath165 implied by the best - fit model , from both the observations and the model .
by contrast to figure [ f : obsvar ] , this now also subtracts the spatially - varying viewing perspective .
so the observed rotation component @xmath166 is compared to the model component @xmath167 .
clockwise motion is clearly evident in the observations , and this is reproduced by the model .
the best - fit model has @xmath168 for @xmath169 . hence , @xmath170 .
so even though the model captures the essence of the observations , it is not formally statistically consistent with it
. there are three possible explanations for this .
first , the observations could be affected by unidentified low - level systematics in the data analysis , in addition to the well - quantified random uncertainties .
second , shot noise from the finite number of stars may be important for some fields with low @xmath18 , causing the mean pm of the observed stars to deviate from the true mean motion in the lmc disk . and third
, the model may be too over - simplified ( e.g. , if there are warps in the disk , or if the streamlines in the lmc disk deviate from circles at a level comparable to our measurement uncertainties ) .
it is difficult to establish which explanation may be correct , and the explanation may be different for different fields .
two of our hst fields are close to each other at a separation of only @xmath171 , and this provides some additional insight into potential sources of error .
the fields , labeled l12 and l14 in table 1 of paper i , are located at @xmath172 and @xmath173 ( see figure [ f : obsvar ] ) . since the fields are so close to each other , the best - fit model predicts that the pms should be similar , @xmath174 mas / yr . however ,
the observations differ by @xmath175 mas / yr .
this level of disagreement can in principle happen by chance ( 9% probability ) , but maybe a possible additional source of error is to blame .
the disagreement in this case can not arise because the model is too oversimplified , since almost any model would predict that closely - separated fields in the disk have similar pms .
also , shot noise is too small to explain the difference .
these fields had @xmath17618 stars measured , and a typical velocity dispersion in the disk is @xmath177 ( vdm02 ) .
this implies a shot noise error ( per coordinate , per field ) of only @xmath178 mas / yr , which is below the random errors for these fields .
these fields have lower @xmath18 and smaller random errors than most other fields , so this means that shot noise in general plays at most a small role .
so in the case of these fields , and maybe for the sample as a whole , it is likely that we are dealing with unidentified low - level systematics in the data analysis . either way , the exact cause why @xmath179 does not matter much , since in the monte - carlo analysis of pseudo - data we multiply all observational errors by @xmath151 .
so the actual residuals in the data - model comparison are accounted for when calculating the uncertainties in the model parameters .
moreover , the astrometric observations presented in paper i are extremely challenging .
so the level of agreement in the data - model comparison of figure [ f : obsrot ] is actually extremely encouraging .
many studies exist of the los velocity field of tracers in the lmc , as discussed in section [ s : intro ] .
two of the most sophisticated studies are those of vdm02 and o11 .
the vdm02 study modeled the los velocities of @xmath180 carbon stars , and its results formed the basis of the rotation model used in k06 .
the more recent o11 study obtained a rotation fit to the los velocities of @xmath181 red supergiants ( rsgs ) , and also presented @xmath182 new los velocities for other giant and agb stars .
the parameters of the vdm02 and o11 rotation models are presented in table [ t : paramlit ] .
comparison of the vdm02 and o11 parameters to those obtained from our pm field fit in table [ t : param ] shows a few important differences .
the com pm values used by both vdm02 and o11 are inconsistent with our most recent estimate from paper i. this is important , because the transverse motion of the lmc introduces a solid body rotation component into the lmc los velocity field , which must be corrected to model the internal lmc rotation .
also , the dynamical centers either inferred ( vdm02 ) or used ( o11 ) by the past los velocity studies are in conflict with the dynamical center implied by the new pm analysis . these differences are discussed in detail in section [ s : param ] .
motivated by these differences , we decided to perform a new analysis of the available los velocity data from the literature , taking into account the new pm results .
this yields a full three - dimensional view of the rotation of the lmc disk .
= 0.8 it is well - known that the kinematics of stars in the lmc depends on the age of the population , as it does in the milky way .
young populations have small velocity dispersions , and high rotation velocities .
by contrast , old populations have higher velocity dispersions ( e.g. , van der marel et al . 2009 ) , and lower rotation velocities ( see table [ s : param ] ) due to asymmetric drift .
for this reason , we compiled two separate samples from the literature for the present analysis : a `` young '' sample and an `` old '' sample .
the young sample is composed of rsgs , which is the youngest stellar population for which detailed accurate kinematical data exist .
the old sample is composed of a mix of carbon stars , agb stars , and rgb stars . for our young sample , we combined the rsg velocities of prevot et al .
( 1985 ) , massey & olsen ( 2003 ) , and o11 ( adopting the classification from their figure 1 ) .
for the old sample , we combined the carbon star velocities of kunkel et al .
( 1997 ) , hardy et al .
( 2001 ; as used also by vdm02 ) , and o11 ; the oxygen - rich and extreme agb star velocities of o11 ; and the rgb star velocities of zhao et al .
( 2003 ; selected from their figure 1 using the color criterion @xmath183 ) , cole et al .
( 2005 ) , and carrera et al .
( 2011 ) .
when a star is found in more than one dataset , we retained only one of the multiple velocity measurements . if a measurement existed from o11 , we retained that , because the o11 data set is the largest and most homogeneous dataset available . otherwise we retained the measurement from the data set with the smallest random errors .
stars with non - conforming velocities were rejected iteratively using outlier rejection .
for the young and old samples we rejected stars with velocities that differ by more than @xmath184 and @xmath185 from the best - fit rotation models , respectively . in each case
this corresponds to residuals @xmath186 , where @xmath187 is the los velocity dispersion of the sample .
the outlier rejection removes both foreground milky way stars , as well as stripped smc stars that are seen in the direction of the lmc ( estimated by o11 as @xmath188% of their sample ) .
all samples were brought to a common velocity scale by applying additive velocity corrections to the data for each sample . these were generally small ; massey & olsen ( 2003 ) : @xmath189 ; kunkel et al .
( 1997 ) : @xmath190 ; hardy et al .
( 2001 ) : @xmath191 ; cole et al .
( 2005 ) : @xmath192 ; carrera et al .
( 2011 ) : @xmath193 . ] , except for the zhao et al .
( 2003 ) sample ; f056 conf 02 : @xmath194 ; f056 conf 04 : @xmath195 ; f056 conf 05 : @xmath196 ; f056 conf 21 : @xmath197 ; fields as defined in table 1 of zhao et al .
( 2003 ) . ] .
we adopted the absolute velocity scale of o11 as the reference .
since they observed both young and old stars in the same fields with the same setup , this ties together the velocity scales of the young and old samples .
to bring other samples to the o11 scale we used stars in common between the samples , and we also compared the residuals relative to a common velocity field fit .
our final samples contain los velocities for 723 young stars and 6067 old stars in the lmc .
figure [ f : losobs ] shows a visual representation of the discrete velocity field defined by the stars in the combined sample . the coverage of the lmc is patchy and incomplete , as defined by the observational setups used by the various studies .
the young star sample is confined almost entirely to distances @xmath198 degrees from the lmc center .
this is where the old star sample has most of its measurements as well .
however , a sparse sampling of old star velocities does continue all the way out to @xmath199 degrees from the lmc center .
a velocity gradient is easily visible in the figure by eye .
what is observed is the sum of the internal rotation of the lmc and an apparent solid - body rotation component due to the lmc s transverse motion ( vdm02 ) . the latter component contributes more as one moves further from the lmc center , which causes an apparent twisting of the velocity field with radius . to interpret the los velocity data we use the same rotation field model for a circular disk as in section [ ss : vmodel ] .
the model is defined by the 7 parameters @xmath200 and the one - dimensional function @xmath61 , which we parameterize with the two parameters @xmath125 and @xmath126 as in equation ( [ rotparam ] ) .
note that the los velocity field depends on the physical velocities @xmath201 , @xmath202 , @xmath94 , and @xmath61 , unlike the pm field , which depends on the angular velocities @xmath86 , @xmath90 , @xmath66 , and @xmath47 .
as before , the model can be written as a sum of two terms , @xmath203 , representing the contributions from the systemic motion of the lmc com and from the internal rotation of the lmc , respectively .
the analytical expressions for the los velocity field @xmath204 thus obtained were presented in vdm02 .
as before , we refer the reader to that paper for the details of the spherical trigonometry and linear algebra involved . in section [ s : model ] we have fit the pm data by themselves , and in other studies such as vdm02 and o11 , the los data have been fit by themselves .
these approaches require that some systemic velocity components ( @xmath94 for the pm field analysis , and @xmath45 for the los velocity field analysis ) must be fixed a priori to literature values .
but clearly , the best way to use the full information content of the data is to fit the pm and los data _
simultaneously_. this is therefore the approach we take here . to fit the combined data , we define a @xmath130 quantity @xmath205 the quantity @xmath133
is as defined in equation ( [ chisqpm ] ) .
the observational pm errors are adjusted as in section [ ss : compare ] so that the best fit to the pm data by themselves yields @xmath206 .
similarly , we define @xmath207 ^ 2 , \label{chisqlos}\ ] ] which sums the squared residuals over all @xmath18 los velocities . here
@xmath208 is a measure of the observed los velocity dispersion of the sample , which we assume to be a constant for each los velocity sample .
we set @xmath208 to be the rms scatter around the best - fit model that is obtained when the los data are fit by themselves ( this yields @xmath209 , analogous to the case for @xmath133 ) .
= 0.8 this approach yields that @xmath210 for the young sample , and @xmath211 for the old sample .
this confirms , as expected , that the older stars have a larger velocity dispersion .
these results are broadly consistent with previous work ( e.g. , vdm02 ; olsen & massey 2007 ) .
note that @xmath212 represents a quadrature sum of the intrinsic velocity dispersion @xmath213 of the stars and the typical observational measurement error @xmath214 .
for all the data used here , @xmath215 , so it is justified to not include the individual measurement errors @xmath216 explicitly in the definition of @xmath217 .
as before , we minimize @xmath130 as function of the model parameters using a down - hill simplex routine ( press et al . 1992 ) , with multiple iterations and checks built in to ensure that a global minimum is found .
we calculate error bars on the best - fit model parameters using monte carlo simulations .
the pseudo pm data for this are generated as in section [ ss : method ] .
the pseudo los velocity data are obtained by drawing new velocities for the observed stars . for this
we use the predictions of the best - fit model , to which we add random gaussian deviates that have the same scatter around the fit as the observed velocities . in minimizing @xmath130
, we treat all model parameters as free parameters that are used to optimize the fit . however , we keep the distance fixed at @xmath162 ( freedman et al . 2001 ) . the uncertainty @xmath218 is accounted for by including it in the monte - carlo simulations that determine the uncertainties on the best - fit parameters .
as discussed later in section [ ss : dist ] , the combination of pm and los data does constrain the distance independently . however
, this does not ( yet ) yield higher accuracy than conventional methods . the stars for which we have measured pms form essentially a magnitude limited sample , composed of a mix of young and old stars .
this mix is expected to have a different rotation velocity than a sample composed entirely of young or old stars .
for this reason , we allow the rotation amplitude @xmath102 in the pm field model to be different from the rotation amplitude @xmath106 in the velocity field model .
both amplitudes are varied independently to determine the best - fit model .
however , we do require the scale length @xmath126 of the rotation curve and also the parameters that determine the orientation and dynamical center of the disk to be the same for the pm and los models . with this methodology
, we do two separate fits .
the first fit is to the combination of the pm data and the young los velocity sample , and the second fit is to the combination of the pm data and the old los velocity sample .
this has the advantage ( compared to a single fit to all the data , with only a different rotation amplitude for each sample ) of providing two distinct answers .
comparison of the results then provides insight into both the systematic accuracy of the methodology , and potential differences in geometrical or kinematical properties between different stellar populations .
table [ t : param ] lists the parameters of the best - fit model and their uncertainties .
the quality of the model fits to the pm data is similar to what was shown already in figure [ f : obsrot ] for fits that did not include any los velocity constraints .
a data - model comparison for the fits to the los velocity data is shown in figure [ f : obslos ] .
the fits are adequate .
it is clear that the young stars rotate more rapidly than the old stars , and have a smaller los velocity dispersion .
the continued increase in the observed rotation amplitude with radius is due to the solid - body rotation component in the observed velocity field that is induced by the transverse motion of the lmc .
the parameters for the best fit models to the combined pm and los velocity samples can be compared to the results obtained when only the pms are fit ( table [ t : param ] ) , or the results that have been obtained in the literature when only the los velocities were fit ( table [ t : paramlit ] ) .
this shows good agreement for some quantities , and interesting differences for others .
we proceed in section [ s : param ] by discussing the results and their comparisons in detail , and what they tell us about the lmc .
= 0.8 the lmc is morphologically peculiar in its central regions , with a pronounced asymmetric bar .
moreover , the light in optical images is dominated by the patchy distribution of young stars and dust extinction . as a result ,
the lmc has become known as a prototype of `` irregular '' galaxies ( e.g. , de vaucouleurs & freeman 1972 ) . however , the old stars that dominate the mass of the lmc show a much more regular large - scale morphology .
this is illustrated in figure [ f : obscontour ] , which shows the number density distribution of red giant and agb stars extracted from the 2mass survey ( vdm01 ) . despite this large - scale regularity
, there does not appear to be a single well - defined center .
it has long been known that different methods and tracers yield centers that are not mutually consistent , as indicated in the figure . the densest point in the lmc bar
is located asymmetrically within the bar , on the south - east side at @xmath219 ( vdm01 ; de vaucouleurs & freeman 1972 ) .. ] the center of the outer isoplets in figure [ f : obscontour ] , corrected for the effect of viewing perspective , is at @xmath220 ( vdm01 ) .
this is on the same side of the bar , but is offset by @xmath221 .
by contrast , the dynamical center of the rotating hi disk of the lmc is on the opposite side of the bar , at @xmath222 ( kim et al .
1998 ; luks & rohlfs 1992 ) .
this is @xmath223 , i.e. , more than 1 kpc , away from the densest point in the bar ( @xmath224 at @xmath225 ) .
these offsets do not pose much of a conundrum .
numerical simulations have established that an asymmetric density distribution and offset bar in the lmc can be plausibly induced by tidal interactions with the smc ( e.g. , bekki et al . 2009 ; besla et al .
what has been more puzzling is the position of the stellar dynamical center at @xmath226 , as determined by vdm02 from the los velocity field of carbon stars .
olsen & massey ( 2007 ) independently fit the same data , and obtained a position ( and other velocity field fit parameters ) consistent with the vdm02 value .
the vdm02 stellar dynamical center was adopted by subsequent studies of los velocities ( e.g. , o11 ) and pms ( k06 , p08 ) , without independently fitting it .
this position is consistent with the densest point of the bar and with the center of the outer isophotes .
but it is @xmath227 away from the hi dynamical center .
vdm02 argued that this may be due to the fact that hi in the lmc is quite disturbed , and may be subject to non - equilibrium gas - dynamical forces .
however , more recent numerical simulations in which the morphology of the lmc is highly disturbed due to interactions with the smc have shown that the dynamical centers of the gas and stars often stay closely aligned ( besla et al . 2012 ) .
the best - fit stellar dynamical center from our model fit to the pm field is at @xmath228 .
this _ agrees _ with the hi dynamical center ( see figure [ f : obscontour ] ) .
but it differs from the stellar dynamical center inferred by vdm02 by @xmath229 , which is inconsistent at the 99% confidence level .
this is surprising , because the pm field and los velocity field are simply different projections of the three - dimensional velocity field of the stellar population .
so one would expect the inferred dynamical centers to be the same .
when we fit the pm data and los velocities simultaneously ( section [ s : los ] ) , we find centers that are somewhat intermediate between between the pm - only dynamical center , and the vdm02 dynamical center ( see figure [ f : obscontour ] ) .
this is a natural outcome , as these model fits try to compromise between datasets that apparently prefer different centers .
the old star sample that we use here is some six times larger than the sample used by vdm02 , and yields a center that is consistent with the young star sample used here .
hence , the fact that los velocities prefer a stellar dynamical center more towards the south - east of the bar is a generic result , and does not appear to be due to some peculiarity with the carbon star sample used by vdm02
. however , the dynamical centers that we infer from the combined pm and los samples are much closer to the hi dynamical center than the vdm02 dynamical center .
specifically , the offsets from the hi center are @xmath230 for the old @xmath231 sample and @xmath232 for the young @xmath231 sample .
such offsets occur by chance only 9% and 6% of the time , respectively .
hence , they most likely signify a systematic effect and not just a chance occurrence . in reality
, it is likely that the hi and stellar dynamical centers are coincident , since both the stars and the gas orbit in the same gravitational potential .
some unknown systematic effect may therefore be affecting the los velocity analyses .
for example , there is good reason to believe that the true dynamical structure of the lmc is more complicated than the circular orbits in a thin plane used by our models ( e.g. , warps and twists of the disk plane have been suggested by vdmc01 , olsen & salyk 2002 , and nikolaev et al .
the uncertainties thus introduced may well affect different tracers differently , leading to systematic offsets such as those reported here .
visual inspection of the pm vector field in figure [ f : obsrot ] strongly supports that the center of rotation must be close to the position identified by the pm - only model fit .
for example ,
the pm vectors in the central region do not have a definite sense of rotation around the position identified by vdm02 .
visual inspection of the los velocity field in figure [ f : obslos ] shows the difficulty of determining an accurate center from such data .
either way , the results in table [ t : param ] and figure [ f : obscontour ] definitely indicate the lmc stellar dynamical center is much closer to the hi dynamical center than was previously believed .
existing constraints on the orientation of the lmc disk come from two techniques .
the first technique is a geometric one , based on variations in relative distance to tracers in different parts of the lmc disk ( vdmc01 ) .
the second is a kinematic method , based on fitting circular orbit models to the velocity field of tracers , as we have done here .
the geometric technique yields both the inclination and line - of - nodes position angle . when applied to los velocities , the kinematic technique yields only the line - of - nodes position angle , since the inclination is degenerate with the amplitude of the rotation curve .
but when applied to pms , the kinematic technique yields both viewing angles ( see section [ ss : pmvslos ] ) .
existing constraints on the disk orientation obtained with these techniques were reviewed in , e.g. , van der marel ( 2006 ) and van der marel et al .
some more recent results have appeared in e.g. koerwer ( 2009 ) , o11 , haschke et al .
( 2012 ) , rubele et al .
( 2012 ) , and subramanian & subramaniam ( 2013 ) .
all studies in the past decade or so agree that the inclination is in the range @xmath233@xmath234 , and that the line - of - nodes position angle is in the range @xmath235@xmath236 .
the variations between the results from different studies are large , and often exceed significantly the random errors in the best - fit parameters
. some of this variation may be real , and due to spatial variations in the viewing angles due to warps and twists of the disk plane , combined with differences in spatial sampling between studies , differences between different tracer populations , and contamination by possible out of plane structures ( e.g. , o11 ) .
our best - fit model to the pm velocity field has @xmath237 and @xmath238 .
the implied viewing geometry of the disk is illustrated in figure [ f : obsrot ] .
the inferred orientation angles are within the range of expectation based on previous work , although they are at the high end .
however , they are perfectly plausible given what is known about the lmc .
this is an important validation of the accuracy of the pm data and of our modeling techniques .
it is the first time that pms have been used to derive the viewing geometry of any galaxy .
however , the random errors in our estimates are not sufficiently small to resolve the questions left open by past work ( apart from the fact that variations in previously reported values appear to be dominated by systematic variations , and not random errors ) .
when we fit not only the pm data , but also los velocities , the best - fit viewing angles change ( table [ t : param ] ) , in some cases by more than the random errors .
however , all inferred values continue to be within the range of what has been reported in the literature . the best - fit inclination with pm data and
the old star @xmath231 sample is @xmath239 , consistent e.g. with the value @xmath240 inferred geometrically by vdmc01 ( and used subsequently in the kinematical studies of vdm02 and o11 ) .
the best - fit line - of - nodes position angle with the pm data and the old star @xmath231 sample is @xmath241 .
this is somewhat larger than the carbon star result @xmath242 obtained by vdm02 , due primarily to the different dynamical center inferred here . the best - fit line - of - nodes position angle with the pm data and the young star @xmath231 sample is @xmath243 .
this is larger than the result @xmath244 obtained by o11 for the same @xmath231 sample , due primarily to the different dynamical center inferred here .
the best - fit inclination with the pm data and the young star @xmath231 sample is @xmath245 .
this is somewhat smaller than , but consistent with , the value obtained when the old star @xmath231 sample is used .
however , the line - of - nodes position angles for the fits with the young and old stars differ by @xmath246 .
this is an intriguing result , since the data for these samples were analyzed in identical fashion , and they do yield consistent dynamical centers .
this suggests that there may be real differences in the disk geometry or kinematics for young and old stars , apart from their rotation amplitudes .
indeed , the values inferred here kinematically using young stars are consistent with the values inferred geometrically for ( young ) cepheids , by nikolaev et al .
they found that @xmath247 and @xmath248 .
by contrast , the values inferred here kinematically using old stars are more consistent with some sets of orientation angles that have been inferred geometrically for agb and rgb stars ( e.g. , vdmc01 ; olsen & salyk 2002 ) .
all results obtained here confirm once again that the position angle of the line of nodes differs from the major axis of the projected lmc body , which is at @xmath249 .
this implies that the lmc is not circular in the disk plane ( vdm01 ) . in the best - fit model to the pm data , the final result for the lmc com pm is @xmath250 mas / yr .
paper i presented a detailed discussion of this newly inferred value , including a comparison to previous hst and ground - based measurements .
there are three components that contribute to the final pm error bars , namely : ( 1 ) the random errors in the measurements of each field ; ( 2 ) the excess scatter between measurements from different fields that is not accounted for by random errors , disk rotation , and viewing perspective ; and ( 3 ) uncertainties in the geometry and dynamics of the best - fitting disk model .
the contribution from the random errors can be calculated simply by calculating the error in the weighted average of all measurements .
this yields @xmath251 mas / yr . this sets an absolute lower limit to how well one could do in estimating the lmc com pm from these data , if there were no other sources of error .
as discussed above , the scatter between fields increases the error bars by a factor @xmath252 .
therefore , @xmath253 mas / yr . since errors add in quadrature , this implies that @xmath254 mas / yr . and finally the contribution from uncertainties in geometry and dynamics of the best - fitting disk model are @xmath255 mas / yr and @xmath256 mas / yr . the final errors bars equal @xmath257 .
so our knowledge of the geometry and kinematics of the lmc disk is now the main limiting factor in our understanding of the pm of the lmc com .
the exact position of the lmc dynamical center is an important uncertainty in models of the lmc disk .
for this reason , we explored explicitly how the fit to the pm velocity field depends on the assumed center .
for example , we ran models in which the center was kept fixed to the position identified by vdm02 ( even though this center is strongly ruled out by our data ) .
this changes only one of the com pm components significantly , namely @xmath90 , the lmc com pm in the north direction .
its value increases by @xmath258 mas / yr when the vdm02 center is used instead of the best - fit pm center .
when we use instead the centers from our combined pm and los velocity fits , then @xmath90 increases by @xmath259@xmath260 mas / yr , while again @xmath86 stays the same to within the uncertainties ( see table [ t : param ] ) .
we have found more generally that if the center is moved roughly in the direction of the position angle of the lmc bar ( pa @xmath261 ; vdm01 ) , then the implied @xmath90 changes while the implied @xmath86 is unaffected . if instead the center is moved roughly perpendicular to the bar , then @xmath86 changes while the implied @xmath90 is unaffected .
as discussed in paper i , @xmath86 affects primarily the galactocentric velocity of the lmc , while @xmath90 affects primarily the direction of the orbit as projected on the sky . in practice ,
all of the centers that have been plausibly identified for the lmc align roughly along the bar ( see figure [ f : obscontour ] ) .
any remaining systematic uncertainties in the lmc center position therefore affect primarily @xmath90 , and not @xmath86 . in our fits to the pm field we kept the parameter @xmath262
mas / yr fixed to the value implied by pre - existing measurements . however , we did also run models in which it was treated as a free parameter . this yielded @xmath263 mas / yr . this is consistent with the existing knowledge , but not competitive with it in terms of accuracy .
interestingly , the result does show at statistical confidence that @xmath68 .
so the observed pm field in figure [ f : obsvar ] contains enough information to demonstrate that the lmc is moving away from us .
this is analogous to the situation for the los velocity field , which contains enough information to demonstrate that the lmc s transverse velocity is predominantly directed westward ( figure 8 of vdm02 ) . in our fits of the combined pm and
los velocity data , we did fit independently for the systemic los velocity . when using the old star @xmath231 sample , this yields @xmath264 .
this is consistent with the results of vdm02 and olsen & massey ( 2007 ) .
however , when using the young star @xmath231 sample , we obtain @xmath265 .
this differs significantly both from the old star result , and from the result of o11 for the same young star sample ( table [ t : paramlit ] ) .
this is a reflection of the different centers used in the various fits , and is not due to an intrinsic offset in systemic velocity between young and old stars .
when we fit the young star data with a center that is fixed to be identical to that for the old stars , we do find systemic velocities @xmath94 that are mutually consistent . in the best - fit model to only the pm data , the rotation curve rises linearly to @xmath266 , and then stays flat at @xmath267 .
at a distance modulus @xmath268 ( freedman et al .
2001 ) , this implies that @xmath269 and @xmath270 .
this rotation curve fit is shown by the black lines in figure [ f : rotcurve ] . to further assess the pm rotation curve
, we also obtained a non - parametric estimate for it .
for each hst field we already have from figure [ f : obsrot ] the observed rotation component @xmath271 , as well as the best - fit model component @xmath272 .
the model also provides the in - plane rotation velocity @xmath273 at the field location .
we then estimate the observed rotation velocity for each field as @xmath274 ( { \vec \mu}_{\rm obs , rot } \cdot { \vec \mu}_{\rm rot } / |{\vec \mu}_{\rm rot}|^2)$ ] , where @xmath275 designates the vector inner product .
this corresponds to modifying the model velocity by the component of the residual pm vector that projects along the local direction of rotation .
similarly , the uncertainty @xmath276 is estimated as the projection of the observational pm error ellipse onto the rotation direction .
figure [ f : rotcurve ] shows the rotation curve thus obtained .
results are shown for the individual hst fields , color - coded as in figures [ f : obsvar ] , [ f : obsrot ] , and [ f : obscontour ] by whether two or three epochs of data are available .
the three - epoch measurements have good accuracy ( median @xmath277 ) .
by contrast , the two - epoch measurements have much larger uncertainties ( median @xmath278 ) , as was the case in p08 .
so for the two - epoch data we also plot the results obtained upon binning in @xmath279 bins of size @xmath280 .
the rotation curve defined by combining the three - epoch and binned two - epoch data is listed in table [ t : rotcurve ] .
the unparameterized rotation curve is fit reasonably well by the simple parameterization given by equation ( [ rotparam ] ) .
rrrrrrl 0.0112 & 0.56 & 0.117 & 0.080 & 27.7 & 19.0 & l7,21 + 0.0118 & 0.59 & 0.159 & 0.023 & 37.8 & 5.6 & l3 + 0.0274 & 1.37 & 0.360 & 0.085 & 85.5 & 20.1 & l5,13,15,19 + 0.0360 & 1.80 & 0.250 & 0.050 & 59.4 & 11.9 & l12 + 0.0365 & 1.83 & 0.360 & 0.060 & 85.6 & 14.3 & l14 + 0.0449 & 2.25 & 0.192 & 0.063 & 45.6 & 15.0 & l8,9,20 + 0.0497 & 2.49 & 0.211 & 0.063 & 50.2 & 15.0 & l4 + 0.0519 & 2.60 & 0.333 & 0.059 & 79.2 & 14.1 & l16 + 0.0623 & 3.12 & 0.246 & 0.108 & 58.5 & 25.6 & l10,17,18 + 0.0693 & 3.47 & 0.308 & 0.045 & 73.2 & 10.6 & l22 + 0.0749 & 3.76 & 0.355 & 0.041 & 84.4 & 9.7 & l1 + 0.0872 & 4.37 & 0.361 & 0.035 & 85.7 & 8.3 & l2 + 0.0886 & 4.44 & 0.481 & 0.070 & 114.4 & 16.6 & l6 + 0.0930 & 4.66 & 0.330 & 0.049 & 78.3 & 11.6 & l11 + = p08 estimated the pm rotation curve from only the two - epoch pm data .
their rotation velocity amplitude @xmath281 was surprisingly high .
this exceeds the value derived from the radial velocities of hi and young stars by approximately 30@xmath282 ( o11 ) .
it would be hard to understand how any stars in the lmc could be rotating significantly faster than the hi gas .
when we use the method discussed above on our own reanalysis of the two - epoch pm data , the resulting unparameterized rotation curve is qualitatively similar to that of p08 , but the uncertainties are very large . with the improved quality of our three - epoch data ,
the rotation curve is much better determined . moreover
, the rotation amplitude comes down to @xmath270 .
this is more in line with expectation , and removes a significant puzzle from the previous work .
the uncertainties in our unparameterized rotation curve in table [ t : rotcurve ] are underestimates , because they do not take into account the propagated uncertainties in other model parameters ( such as the dynamical center , viewing angles , and com motion ) .
the weighted average @xmath283 in table [ t : rotcurve ] for @xmath284 equals @xmath285 mas / yr . by contrast , the best - fit from the parameterized model in table [ t : param ] is @xmath99 mas / yr . since errors add in quadrature , the uncertainties in the other model parameters contribute an uncertainty of @xmath286 mas / yr to the rotation amplitude .
this dominates the error budget , even though it is small compared to the typical per - field error bars in table [ t : rotcurve ] .
so the per - field pm uncertainties are not the main limiting factor in our understanding of the rotation curve amplitude .
when we fit not only the pm data , but also the los velocity data , then the pm rotation amplitude @xmath102 changes by about the random uncertainty @xmath287 .
when we fit the old star @xmath231 sample , @xmath102 goes up , and when we fit the young star sample , @xmath102 goes down .
this is because the inclusion of the los velocities alters the best - fit line - of - nodes position angle @xmath44 to lower or higher values , respectively ( table [ t : param ] ) .
this affects @xmath102 , because our monte - carlo simulations show that @xmath44 is anti - correlated with @xmath125 . in our best - fit models to the @xmath231 data ( combined with the pm data ) ,
the rotation amplitude @xmath106 is not very accurately determined .
this is because only a fraction @xmath288 of any velocity is observed along the line of sight .
the inclination is not accurately known from our or any other data ( see section [ ss : orient ] ) , and the deprojection therefore introduces significant uncertainty .
by contrast , @xmath110 is determined much more accurately . for our old star sample
, we find that @xmath289 .
this is consistent with the result from vdm02 ( see table [ t : paramlit ] ) . for our young star sample , we find that @xmath290 .
so the young stars have a higher rotation curve amplitude than the old stars , consistent with previous findings .
however , the value inferred here is about 20% less than the value implied by the rotation curve fits of o11 , for the same sample of stars ( but not including pms ) . this is due primarily to the larger value of @xmath44 inferred here . as for the pm case
, we also determined unparameterized rotation curves from the los velocity data , separately for the young and old star samples . for this we kept all model parameters fixed , except the rotation amplitude , to the values in table [ t : param ] .
we then binned the stars by their radius @xmath291 in the disk , and fit the rotation amplitude separately for each radial bin .
the rotation curves thus obtained are listed in table [ t : rotcurvel ] .
the uncertainties only take into account the shot noise from the finite number of stars .
this yields underestimates , because it does not take into account the propagated uncertainties in other model parameters .
the inferred rotation curves are shown in figure [ f : rotcurvel ] , together with the parametrized fits from table [ t : param ] .
the rotation curves are well fit by the simple parameterization given by equation ( [ rotparam ] ) . for the parameterized fits , the uncertainty in the amplitude shown is @xmath292
; so this includes the propagated uncertainty from all model parameters except the inclination . in general , for all rotation curve results derived here from los velocities , the inclination is the dominant uncertainty ( @xmath293 \pi / \tan i$ ] ) .
the turnover radii in our rotation curve fits , @xmath294 for the old stars , and @xmath295 for the young stars , are similar to what was found by vdm02 and o11 , respectively ( table [ t : paramlit ] ) .
the value of @xmath126 for the young stars is only about half that for the old stars .
so the young stars not only have a higher rotation curve amplitude , but the rotation curve also rises faster .
the value of the turnover radius @xmath113 inferred from the fit to only the pm data , @xmath296 , is even lower than the value for the young star @xmath297 sample , but only at the @xmath298 level .
we do not attach much significance to this , given the sparse radial sampling of our pm data , especially with high - quality wfc3 fields at small radii ( only one field at @xmath299 kpc ; see figure [ f : rotcurvel ] ) . the radial behavior and turnover of the rotation curve
are therefore more reliably constrained by the los data than by the pm data .
the values of @xmath106 implied by our fits are @xmath300 for the old stars , and @xmath109 for the young stars , respectively .
these results are consistent with the results obtained by vdm02 and o11 ( table [ t : paramlit ] ) .
it should be noted that while o11 reported @xmath301 for the same sample of young stars , their listed uncertainty did not include the uncertainty from propagation of uncertainties in the center , inclination , com pm , or distance .
the inclination alone ( from vdmc01 , as adopted by o11 ) adds a @xmath302 uncertainty .
so while the random uncertainties between our fit and that of o11 are in fact similar , our result should be more accurate in a systematic sense .
this is because of our new determination of , e.g. , the dynamical center and the com pm .
the good agreement between the @xmath106 values reported here and in o11 is actually somewhat fortuitous .
we find the los component of the rotation to be @xmath303% less than o11 did , but they adopted a larger inclination .
= the rotation amplitude inferred from our pm data , @xmath304 , falls between the values inferred from the los velocities of old stars , @xmath305 , and young stars , @xmath306 , respectively ( see also figure [ f : rotcurvel ] ) .
this is because our stellar pm sample is essentially a magnitude limited sample , composed of a mix of young and old stars . to assess quantitatively whether the rotation amplitudes derived from the pm data and los velocities are consistent ,
let us assume that a fraction @xmath307 of the stars that contribute to our pm measurements are young , and a fraction @xmath308 are old .
the designations `` young '' and `` old '' in this context refer to the fact that the stars are assumed to have the same kinematics as the stars in our young and old @xmath297 samples .
this implies an expected pm - inferred rotation amplitude @xmath309 ^ 2 + [ f18.8]^2 ) { \ { \rm{km \ s^{-1}}}\>}$ ] . equating this with the observed
@xmath102 implies that @xmath310 .
cccccc 0.011 & 0.5 & 11.2 & 4.7 & 10.6 & 1.4 + 0.026 & 1.3 & 54.9 & 3.1 & 20.3 & 1.4 + 0.044 & 2.2 & 90.7 & 2.4 & 29.2 & 1.6 + 0.060 & 3.0 & 89.0 & 3.6 & 43.7 & 1.7 + 0.076 & 3.8 & @xmath107 & @xmath107 & 62.7 & 2.5 + 0.097 & 4.9 & @xmath107 & @xmath107 & 55.2 & 5.5 + 0.113 & 5.7 & @xmath107 & @xmath107 & 57.4 & 5.4 + 0.131 & 6.6 & @xmath107 & @xmath107 & 34.4 & 6.8 + 0.151 & 7.6 & @xmath107 & @xmath107 & 52.7 & 7.6 + 0.173 & 8.7 & @xmath107 & @xmath107 & 48.3 & 7.6 + figure 6 of k06 shows a color - magnitude diagram ( cmd ) of the lmc stars that contribute to our pm measurements . at the magnitudes of interest ,
there are two main features in this diagram .
there is a blue plume , consisting of main sequence stars and evolved massive stars at the blue edge of their blue loops .
and there is a red plume , consisting mostly of rgb stars and some agb stars .
bright rsgs , such as those in the @xmath231 samples , are too rare to contribute significantly to our small hst fields . to count the relative numbers of blue and red stars , we adopt a separation at @xmath311 .
we then find that the fraction of blue stars ( as ratio of the total stars that contribute to our pm measurements ) increases from @xmath312% at the brightest magnitudes to @xmath313% at the faintest magnitudes .
if we assume that the blue stars have kinematics typical of young stars , and the red stars have kinematics typical of old stars , then this implies @xmath314 .
this cmd - based value is in excellent agreement with the value inferred above from the observed kinematics .
so to within the uncertainties , the observed rotation of the lmc in the pm direction is consistent with the observed rotation in the los direction .
the lmc rotation amplitude @xmath102 inferred from the pm field is relatively insensitive to the inclination ( see section [ ss : pmvslos ] ) .
by contrast , the los velocity data accurately constrain @xmath110 . since the fraction @xmath307 _ must _ be between 0 and 1 , comparison of these quantities can set limits on the inclination
. the inclination must be such that @xmath315_{\rm old } / \sin i \leq v_{0,{\rm pm } } \leq [ v_{0,{\rm los}}\>\sin i]_{\rm young } / \sin i$ ] . with the inferred values from table [ t : param ]
this implies that at @xmath150 confidence @xmath316 . as discussed in section
[ ss : orient ] , this encompasses most of the results reported in the literature .
alternatively , we could be less conservative and assume that we know from the cmd analysis that @xmath317 . in that case
we obtain the more stringent range @xmath318 . but
this assumes that we know the difference in kinematics between different stars in our hst cmds , which has not actually been measured .
so far , we have assumed that the distance @xmath39 to the lmc center of mass is known .
however , a comparison of the pm and los velocity fields does in fact constrain the distance independently , since pms are measured in mas / yr , and los velocities are measured in km / s . as we will discuss , this comparison provides several independent distance constraints .
the first distance constraint is obtained by requiring that the rotation amplitude measured from pms matches that obtained from los velocities .
this is called the `` rotational parallax '' method .
based on the discussion in the previous section , this implies that @xmath319_{\rm young } + ( 1-f ) [ v_{0,{\rm los}}]_{\rm old } ) / [ v_{0,{\rm pm } } / d_0 ] .\ ] ] to use this equation , we must assume the relative fractions of young and old stars that contribute to the pm measurements .
using the analysis in section [ sss : rotcomp ] , we set @xmath320 . with the inferred values from table [ t : param ]
this implies that @xmath321 .
this is consistent with existing knowledge ( e.g. , freedman et al . 2001 ) .
however , the uncertainty is very large , due primarily to the uncertainties in the lmc inclination . to obtain a distance estimate with a random error @xmath322
, the inclination would have to be known to better than @xmath323 , not even accounting for other uncertainties .
based on the discussion in section [ ss : orient ] , it is clear that this is not currently the case , despite many papers devoted to the subject .
moreover , one would need to know the fraction @xmath307 more accurately than is possible with only cmd information .
so for the lmc , the method of rotational parallax is not likely to soon yield a competitive distance estimate .
an alternative method to constrain the lmc distance from comparison of the pm and los velocity fields uses the observed los velocities perpendicular to the line of nodes .
rotation is perpendicular to the line of sight there , so that the observed velocities are due entirely to the solid - body rotation induced by the lmc s transverse motion .
hence , the velocities obey @xmath324 , where @xmath28 is the distance from the com , and @xmath325 is the component of the com pm that lies along the line of nodes ( vdm02 ) .
since @xmath325 is constrained by the pm data in mas / yr , the distance @xmath39 can be determined from the los data in km / s . for accurate results ,
this method benefits from having data that extends to large distances @xmath28 , and from having a sample with many velocity measurements . we therefore apply it to the old star @xmath297 sample ( see figure [ f : obslos ] ) . to use the full information content of the data , and to adequately propagate all uncertainties , one must fit the combined pm and old star @xmath231 sample with @xmath326 as a free parameter .
when we do this while keeping @xmath327 fixed to the previously obtained value from table [ t : param ] , we obtain @xmath328 ( similar to an earlier estimate in van der marel et al .
( 2009 ) , which was based on the vdm02 carbon star los velocity data and the k06 com pm estimate ) .
this has a smaller random error than the result from the rotational parallax method , but is still not competitive with existing knowledge .
moreover , it may be a biased estimate .
when fitting @xmath326 , one should really fit @xmath44 simultaneously , because @xmath44 and @xmath326 are generally anti - correlated in our model fits .
however , we found that the multi - dimensional solution space becomes more degenerate when both @xmath44 and @xmath326 are left to vary .
specifically , the best fit @xmath326 can vary by @xmath329 , depending on how we choose to weight the pm data relative to the los velocity data in the @xmath130 definition ( eq . [ [ chitot ] ] ) .
so this method does not currently yield a competitive distance either . a final method for estimating the the lmc distance from comparison of the pm and los velocity fields uses the observed systemic los velocity . as stated in section [ ss :
lmccomlos ] , our pm field fit constrains @xmath330 mas / yr . using the known systemic los velocity @xmath331 for the old star @xmath231 sample , this yields an estimate for the distance : @xmath332 .
again , this is consistent with existing knowledge , but not competitive in terms of accuracy .
the preceding analysis in this paper has assumed that the viewing angles of the lmc disk are constant with time .
vdm02 showed that there are additional contributions to pm and los velocity fields when the viewing angles vary with time , i.e. , @xmath333 or @xmath334 .
this corresponds to a precession or nutation of the spin axis of the lmc disk . to induce
such motion requires external tidal torques .
while it is not impossible that such motion may exist , there is no theoretical requirement that it should .
the main impact of a value @xmath333 is to induce a solid - body rotation component in the los velocity field , with its steepest gradient perpendicular to the line of nodes . to assess the existence of such a component
, we repeated our fits to the combined pm data and old star @xmath231 sample , but now with @xmath335 free to vary .
this yields @xmath336 mas / yr . this result is consistent with zero .
so with the presently available data , there is no need to invoke a non - zero value of @xmath337 .
constraints on @xmath335 from the young star @xmath231 sample are weaker , because those data do nt extended as far from the com , and do nt have as many velocity measurements .
o11 inferred @xmath338 mas / yr for that same sample ( @xmath339 gyr@xmath340 ) .
however , their uncertainty is an underestimate , because it does not propagate the known uncertainties in the center , inclination , com pm , or distance .
based on our analysis of the young stars , we have found no compelling reason to assume they require @xmath341 .
it would in fact be difficult to understand how the spin axis of the young star disk could be moving relative to the old star disk .
a value @xmath342 does not affect the los velocity field .
however , it does cause circular motion in the observed pm field .
this is almost entirely degenerate with the actual rotation of the lmc disk , as measured by the rotation amplitude @xmath102 ( compare figure [ f : obsrot ] ) .
we have shown in section [ sss : rotcomp ] that the amplitude inferred assuming @xmath343 agrees with the rotation amplitudes inferred from los velocities .
therefore , the data do not require a value @xmath334 .
if there is a deviation from zero , it would have to be small enough to not perturb the agreement discussed in section [ sss : rotcomp ] . to estimate the dynamical mass of the lmc , it is necessary to know the kinematics of tracers at large radii .
the outermost tracers for which kinematics are available are the old stars in our @xmath344 sample ( see figure [ f : obslos ] ) , most of which are carbon stars from kunkel et al .
( 1997 ; see figure [ f : losobs ] ) .
figure [ f : rotcurvel ] shows that the rotation curve of these stars stays more - or - less flat out the last data point , at radius @xmath345 in the disk ( see also figure 6 of vdm02 ) .
since this is true for the old stars , it must be true for the young stars as well .
after all , both orbit in the same gravitational potential . based on this reasoning , we infer that the young stars have a rotation amplitude @xmath306 at @xmath345 .
olsen & massey ( 2007 ) inferred a velocity dispersion for these stars of @xmath346 .
the formalism of vdm02 then implies an upward asymmetric drift correction of only @xmath347 , much smaller than the random errors .
this is as expected , given that o11 found that the young stars and hi gas have essentially the same rotation curve .
so we obtain that @xmath348 at @xmath349 kpc , with the error dominated by inclination uncertainties .
the total mass of the lmc inside the last measured data point is @xmath350 , where the gravitational constant @xmath351 .
this yields @xmath352 .
this is consistent with the total mass of the lmc that has been used in several past @xmath18-body simulation studies of the lmc ( e.g. , gardiner & noguchi 1996 ) .
however , this is likely an underestimate of the total lmc mass .
since the rotation curve is flat , the mass likely continues to rise almost linearly beyond @xmath353 .
many virial masses are possible , as long as the concentration of the dark halo is varied to reproduce the dynamical mass . in paper
i we considered models with virial masses up to @xmath354 .
the tidal radius of the lmc can be estimated using the formalism of vdm02 .
we assume that the lmc rotation curve is flat at @xmath355 , and the milky way rotation curve at the distance of the lmc is flat at @xmath356 ( based on the enclosed mass out to the lmc distance given by kochanek 1996 ) .
this implies a tidal radius of @xmath2 ( i.e. , a radius on the sky of @xmath3 ) .
if instead there is no lmc mass outside of @xmath353 , then the tidal radius is smaller by a factor @xmath357 .
either way , the lmc tidal radius is beyond @xmath358 . indeed , photometric studies of the lmc have traced the lmc disk almost this far out ( saha et al .
2010 ) . to determine whether the rotation curve of the lmc is typical
, it is useful to assess how its circular velocity compares to that of other similar galaxies .
the classical tully - fisher relation in spirals has been shown to extend into the low - mass regime when the total baryonic content of the galaxies is used ( gas in addition to stars ; mcgaugh et al .
2005 ; stark et al .
2009 ) . in figure
[ f : btf ] ( blue triangle ) we place the lmc on the baryonic tully - fisher ( btf ) , using @xmath359 from section [ ss : mass ] and @xmath360 from vdm02 .
= mcgaugh ( 2012 ) recently calibrated the btf relation using a sample of gas - dominated low - mass systems , arguing that the errors introduced by modeling the stellar component is minimized in these gas - rich systems .
these galaxies follow a very tight relation . as with high - mass galaxies
, the scatter is below what is expected from initial conditions in cosmological models , implying the need for some feedback process that is correlated with the galaxy potential ( eisenstein & loeb 1996 ; mcgaugh 2012 ) .
similarly , oh et al .
( 2011a ) placed a sample of dwarf galaxies from the hi nearby galaxy survey ( things ; walter et al .
2008 ) , a high velocity - resolution hi survey of dwarf galaxies , on the btf relation . in figure
[ f : btf ] , black points show gas - rich low - mass galaxies compiled in mcgaugh ( 2012 ) and pink diamonds show the things dwarfs from oh et al .
( 2011a , b ) .
the lmc falls on the btf relation followed by these galaxies .
so even though the lmc is a member of an interacting pair , it is not atypical in terms of its btf position .
as discussed in section 4.8 , the error in our @xmath361 is dominated by inclination uncertainties .
so if one assumes _ a priori _ that the lmc must fall exactly on the btf , then this would in principle provide an alternative method to constrain the lmc inclination ( see also oh et al .
also , central density profile slopes are a powerful probe of the properties of dark matter ( navarro , frenk , & white et al . 1997 ; dalcanton & hogan 2001 ) , and these can be constrained from observed rotation curves of dwarf galaxies ( e.g. , de blok 2010 , and references therein ; oh et al .
2011a , b ) . however ,
for the lmc only sparse discrete pm and los datasets are available , and these are not ideally suited for constraining the central rotation curve slope .
also , the possibility of non - circular orbits in the poorly understood lmc bar region would complicate any interpretation .
so we have decided not to pursue a detailed rotation curve decomposition here .
we have presented a detailed study of the large - scale rotation of the lmc based on observed stellar velocities in all three cartesian directions .
this is the first time that such a study has been possible for any galaxy , made possible by the exquisite capabilities of hst for measuring pms in the nearby universe .
while the los velocity field of the lmc has been studied previously using many tracers , our analysis of the pm rotation field is new .
this is important , because the pm rotation field is defined by two components of motion , and it therefore has a higher information content than the los velocity field . as a result , quantities that are degenerate in analyses of the los velocity field ( such as the rotation curve , the inclination , and one component of the transverse motion of the com ) are uniquely determined by analysis of the pm rotation field .
we interpret the data with simple models of circular rotation in a flat disk , which fit the data reasonably well . by and large , we find that the lmc rotation properties as revealed by pm and los data are mutually consistent , as they should be . however , by analyzing accurate pm data and combining it with existing los data we do obtain several new insights into the geometry , kinematics , and structure of the lmc .
previous studies of the lmc have found that the photometric center is offset significantly from the dynamical center defined by the rotating hi gas disk .
this is not difficult to explain , since the lmc has a lopsided off - center bar that could be a transient feature induced by the lmc s interaction with the smc .
what is more puzzling has been the finding that the los velocity field of stellar tracers is best fit by a dynamical center that is _ also _ offset from the hi dynamical center .
our new analysis of the pm rotation field does not confirm this .
we find that the stellar dynamical center revealed by pms _ agrees _ with the hi dynamical center .
however , we also find that previous analyses of the los velocity field were not in error .
our new analysis of the now very large sample of available los velocity data continues to indicate a dynamical center offset , albeit by a smaller amount than reported previously .
this can not be real , since the pm and los analyses observe the same stellar populations .
this likely reveals limitations of the simple rotation model used . in reality
, the stars and gas in the lmc probably do have the same dynamical center , because they orbit in the same gravitational potential .
the best - fit values for the viewing angles that define the orientation of the lmc disk , as inferred from the pm rotation field , are within the range of values implied by previous studies . however , several puzzles remain .
first , the position angle of the line of nodes @xmath44 is not the same for the young and old stellar populations of the lmc .
when los velocities of the young population are fit jointly with the pm data , we obtain @xmath362 . by contrast ,
when los velocities of the old population are fit jointly with the pm data , we obtain @xmath327 .
when the pm data are fit by themselves , the intermediate result @xmath238 is obtained .
the second puzzle is that all these kinematically determined values are larger than several results obtained from geometrical methods ( e.g. , vdmc01 , rubele et al .
similarly for the inclination , the results obtained here and those discussed in the literature span a much larger range than the random errors in the individual measurements .
these results can be explained if the structure of the lmc is more complicated than a single flat disk in circular rotation .
indeed , the data provide indications for variations with both stellar population and radius in the disk . however , by contrast to previous authors we have found no evidence for precession or nutation of the lmc disk . given the latest insights into the position and motion of the lmc com
, we find that acceptable fits to all the kinematical data can be obtained with @xmath363 and @xmath364 .
the lmc rotation curve as implied by our pm measurements has an amplitude @xmath270 .
this applies to a magnitude - limited sample , composed of a mix of stellar populations .
this value of @xmath102 falls between the rotation amplitudes implied by the los velocities of old and young stars , @xmath365 and @xmath109 , respectively .
these results are quantitatively consistent with the natural hypothesis that the blue stars in our hst cmds have predominantly young - star kinematics and the red stars have predominantly old - star kinematics . these results resolve a puzzle posed by analysis of the two - epoch pm rotation curve amplitude .
p08 previously reported @xmath366 , which was difficult to understand as it exceeded the rotation amplitude of both the young stars and the hi gas .
after correction for asymmetric drift , we infer a circular velocity for the lmc of @xmath0 . the large uncertainty is due mostly to uncertainties in the inclination of the lmc .
this @xmath361 places the lmc on the same baryonic tully - fisher relation defined by samples of other low - mass gas - rich galaxies .
also , it implies an enclosed mass @xmath367 , out to the radius @xmath353 to which it has been verified that the rotation curve remains approximately flat .
the virial mass of the lmc should be larger than this , with the exact value depending on how far the lmc s dark halo extends , and on whether it has been tidally truncated . if the lmc circular velocity curve remains flat outside of the region probed observationally , then the tidal radius is @xmath2 ( i.e. , a radius on the sky of @xmath3 ) .
we have discussed three independent methods for combining pm and los velocity information to obtain a kinematical estimate for the lmc distance . while each of these methods yields results that are consistent with existing knowledge , none of them is currently competitive in terms of accuracy .
this is due in large part to the fact that the exact viewing angles of the lmc continue to be poorly understood .
these distance determination methods might become competitive in the future , if better pm data become available and our understanding of the structure and orientation of the lmc improve further .
support for this work was provided by nasa through grants for program go-11730 from the space telescope science institute ( stsci ) , which is operated by the association of universities for research in astronomy ( aura ) , inc .
, under nasa contract nas5 - 26555 .
the authors are grateful to gurtina besla , jay anderson , and charles alcock for contributing to the original observing proposals and to other papers in this series .
knut olsen kindly made available in electronic format the data presented in o11 .
nk would like to thank marla geha s research group for valuable discussions .
|
we present the first detailed assessment of the large - scale rotation of any galaxy based on full three - dimensional velocity measurements .
we do this for the large magellanic cloud ( lmc ) by combining our hubble space telescope average proper motion ( pm ) measurements for stars in 22 fields , with existing line - of - sight ( los ) velocity measurements for 6790 individual stars .
we interpret these data with a model of circular rotation in a flat disk .
the pm and los data paint a consistent picture of the lmc rotation , and their combination yields several new insights .
the pm data imply a stellar dynamical center that coincides with the hi dynamical center ( but offset from the photometric center ) , and a rotation curve amplitude that is consistent with that inferred from los velocity studies .
this resolves several puzzles posed by existing work .
the implied viewing angles of the lmc disk agree with the range of values found in the literature , but continue to indicate variations with stellar population and/or radius in the disk .
young ( red supergiant ) stars rotate faster than old ( red and asymptotic giant branch ) stars due to asymmetric drift . outside the central region ,
the rotation curve is approximately flat out to the outermost data .
the circular velocity @xmath0 ( with the uncertainty dominated by inclination uncertainties ) is consistent with the baryonic tully - fisher relation , and implies an enclosed mass @xmath1 .
the virial mass is larger , depending of the full extent of the lmc s dark halo .
the tidal radius is @xmath2 ( @xmath3 ) , if the circular velocity stays flat this far out .
combination of the pm and los data yields kinematic distance estimates for the lmc , but these are not yet competitive with other methods .
|
since 1995 and the discovery of the first exoplanet orbiting a sun - like star @xcite , the number of detected exoplanets has been steadily increasing . with now almost 1500 confirmed exoplanets ( http://exoplanets.org/ ) and around 20 of them good candidates to host surface liquid water ( http://phl.upr.edu/projects/habitable-exoplanets-catalog ) , we enter in an fascinating age . with the improvements made in exoplanet detection , we are able to detect planets less and less massive ( or smaller and smaller ) approaching the mass ( or size ) range of the earth .
we are now also able to probe the habitable zone of stars .
we define here the habitable zone ( hz ) as the region around a star where a planet with the right atmosphere can potentially sustain surface liquid water @xcite .
most of the detected hz planets are either too massive ( radial velocity planets ) or too large ( transit planets ) to be categorized unequivocally as terrestrial planets .
for example , kepler-22b , a 2.4 @xmath0 planet could be either a mini - neptune or a super - earth @xcite .
however @xcite reports the discovery of the first earth - sized planet in the hz of a low mass star : this planet has a radius only 10% bigger than earth s .
this planet is the closest we know to earth , it is very likely rocky and can potentially host surface liquid water .
unfortunately , being in the hz of a star does not mean that the planet hosts surface liquid water .
the presence of water depends on many different physical parameters and quantities . not only
does it depend on the characteristics of the atmosphere ( pressure , temperature ) and of its composition but also on the dynamics of the orbit of the planet .
indeed , the eccentricity of the orbit as well as the rotation period of the planet and its obliquity ( the angle between the rotation axis and angular momentum vector ) have an effect on the climate of a planet ( the so - called milankovitch cycles that govern the glaciation periods on earth ; * ? ? ?
as all these quantities are influenced by tidal forces , one should take into account the tidal orbital evolution of the system in order to assess the potential of a planet to host water . in section [ tides ]
, we describe how tides do influence the parameters affecting the climate of a planet . in section [ bds ]
, we study the evolution of planets around brown dwarfs which constitutes an extreme case of how dynamics and tides can impact a planet s climate . in section [ k186 ]
, we study kepler-186f , the first earth - sized planet in the hz of a cool star and discuss what can be said about its spin state .
finally , we conclude in section [ conclusion ] .
the tidal model used here is a re - derivation of the equilibrium tide model of @xcite as done in @xcite .
we use the constant time lag model and consider non - coplanar situations as in @xcite .
let us consider a system with n planets orbiting a star .
in order to compute the correct tidal evolution of the system , one has to consider the tide raised by the star on each planet ( called the planetary tide ) but also the tide raised by each planet on the star ( called the stellar tide ) . the planetary tide and stellar tide act on different quantities . depending on the body considered and on its tidal dissipation factor ( a quantity linked to the constant time lag ) , the tides act on different timescales .
the force created by the stellar tide will influence the semi - major axis , eccentricity and inclination of the orbit of the planet .
the resulting torque will impact the rotation rate of the star .
as the mass of the star is usually much higher than that of the planet , the planet can not distort the star significantly and the corresponding evolutions will occur on long timescales .
the semi - major axis evolution of the planet depends on the position of the planet with respect to the corotation radius , which is the orbital radius where the orbital period matches the star s rotation period .
if the planet is inside the corotation radius , the planet migrates inwards and eventually falls on the star .
if the planet is outside the corotation radius , it migrates outwards .
the planet will affect the rotation rate of the star , however this effect is negligible in most cases .
however , the rotation of a star can be significantly increased when a massive planet falls on it @xcite .
unless the star is a very fast rotator , the stellar tide acts to decrease the eccentricity .
when the star is a very fast rotator , a slingshot effect can lead to an increase of the eccentricity .
the stellar tide acts to decrease the inclination of the planet : it brings back the orbital plane into the equatorial plane of the star .
the force created by the planetary tide will influence the semi - major axis and eccentricity of the orbit of the planet .
the resulting torque will impact the rotation rate of the planet : the rotation period and the obliquity .
the star can distort the planet significantly and in particular the evolution of the spin will occur on short timescales . very quickly in the evolution of a circular orbit planet , the rotation will tend towards synchronization , which is a state for which the rotation period of the planet is equal to its orbital period .
if the orbit is eccentric , the planet quickly reaches pseudo - synchronization , which means that its rotation tends to be synchronized with the orbital angular velocity at periastron @xcite . the resulting rotation period is shorter than the orbital period . in the meantime
, the obliquity of the planet quickly tends to zero . on longer timescales ,
after the pseudo - synchronization state is reached , the planetary tide makes the eccentricity and semi - major axis of the planet decrease .
planets around brown dwarfs are interesting to study for three reasons : the hz is located close - in @xcite , so planets in the hz experience strong tidal interactions and these hz planets are easy to detect @xcite .
@xcite investigated the influence of tides on the orbital evolution of single - planet systems orbiting brown dwarfs ( bds ) of different masses .
the bd tide , the tide created by the planet in the bd , makes the planets either fall on the bd or migrate slightly outwards ( see figure [ fig1 ] ) .
not taking this migration into account would over - evaluate the time a planet spends in the hz ( as was done in * ? ? ?
however @xcite showed that , despite this outward migration , planets around bds more massive than @xmath1 could stay in the hz up to a few gigayears . .
the colored lines correspond to the evolution for different initial orbital distances .
the black dashed dotted line corresponds to the corotation radius , the dotted line corresponds to the bd s radius .
the insolation hz is also plotted ( blue region ) .
the thick dashed line corresponds to the roche limit .
@xmath2 is the initial time , named `` time zero '' .
it corresponds to the time of protoplanetary disk dispersal , taken here as 1 myr @xcite.,scaledwidth=80.0% ] let us consider a single planet of @xmath3 orbiting a @xmath1 bd at 0.01 au .
it would reach the hz about @xmath4 myr after the protoplanetary disk dispersal @xcite .
a hundred million years is enough to erode the obliquity of the planet and bring it in pseudo - synchronization .
if the eccentricity is very small the pseudo - synchronization rate is very similar to the synchronization rate ( a difference of @xmath5% for an eccentricity of @xmath6 and of @xmath7% for an eccentricity of @xmath8 ) .
therefore , when the planet reaches the hz , it is likely to have a null obliquity and a synchronous rotation .
this means that this planet always shows the same side to the bd and that its poles receive very little light .
this raises the problem of the possible existence of cold traps : i.e. regions on the planet where the temperature is constantly lower than @xmath9 k and where all the water of the planet will condense @xcite . in this configuration , night side and poles
could be cold traps and although the planet is in the hz , it would not be able to host surface liquid water . however , this situation is not completely hopeless , a dense enough atmosphere that would allow the repartition of heat around the whole planet could prevent cold traps from capturing the water content of the planet .
@xcite showed that gliese 581 d ( if it exists ; * ? ? ?
* ) could have a habitable climate despite a synchronous rotation state .
the situation differs if the planet is part of a multiple planet system . because of planet - planet interactions , eccentricity and obliquity do not tend to 0 but an equilibrium value which is the result of the competition between tidal damping and planet - planet excitation .
an extreme case of planet - planet excitation is the mean motion resonance ( mmr ) .
when two planets are in mmr , the ratio between their orbital periods is commensurable and the eccentricity is excited to higher levels . a close - by example is the 1:2:4 mmr between the 3 inner satellites of jupiter ( io , europa , ganymede ) .
this eccentricity excitation maintained by the resonance causes io to experience an intense internal heating due to the stress it experiences on one orbit .
this internal heating is estimated at about @xmath10 w / m@xmath11 ( e.g. , * ? ? ?
* ) , which is about 40 times higher than the internal heat flux of earth ( about @xmath12 w / m@xmath11 and mainly due to radioactivity ; e.g. , * ? ? ?
* ) . on io
this heat flux gives rise to intense volcanic activity .
a tidally evolving planet in the hz could be heated up and driven in a runaway greenhouse state ( thus creating a tidal venus ; * ? ? ?
let us consider the case of three earth - sized planets orbiting just outside the corotation radius of a @xmath13 bd .
we consider a ) a bd with a tidal dissipation factor similar to hot jupiters ( due to their likeliness in structure , we consider here that bds and hot jupiters have a similar dissipation factor * ? ? ?
* ) and b ) a bd with a tidal dissipation of ten times the dissipation of hot jupiters . in both cases ,
the planets experience a convergent outward migration but they will do so quicker in case b ) than a ) . the difference in dissipation will lead to two dynamically different systems . as the migration is faster in case
b ) planets enter a mmr chain ( 1:2:4 ) after a few million years of evolution .
however in case a ) , they stay outside of resonance .
figure [ fig2 ] shows the evolution of such a system at an older age ( 1 gyr ) , the hz has shrunk and the two inner planets are now in the hz .
the eccentricities of the planets in case a ) are relatively small @xmath14 but in case b ) due to the mmr excitation , they can reach @xmath15 .
in a ) a non - resonant configuration , b ) a resonant configuration ( 1:2 mmr ) .
top graph : the full colored lines correspond to the semi - major axis evolution of the 3 planets , and the dashed lines correspond to their perihelion and aphelion distances .
the blue shaded region is the hz .
middle graph : eccentricity of the 3 planets .
bottom graph : the full colored lines correspond to the tidal heat flux of the 3 planets .
the black dashed dotted line corresponds to @xmath16 w / m@xmath11 , the dashed line corresponds to io s heat flux and the dashed 3 dots line corresponds to earth s heat flux .
the shaded red region corresponds to where the heat flux is so high that the planet is in a runaway greenhouse state.,scaledwidth=90.0% ] figure [ fig2 ] also shows the evolution of the tidal heat flux for each planet , compared to three values : the limit of runaway greenhouse @xmath16 w / m@xmath11 @xcite by @xcite ] , io s tidal heat flux and earth s heat flux . in case
a ) , the average tidal heat flux of the inner planet ( red ) is below the limit of @xmath16 w / m@xmath11 and sometimes exceeds this limit for a few @xmath17 yr . if we consider that the tidal heat flux is the only heat source of the atmosphere of the planet ( we neglect momentarily the insolation from the bd ) , we could say that most of the time the inner planet can have habitable conditions . let us consider that this planet has oceans .
when the tidal heat flux exceeds the runaway greenhouse limit , the oceans will start evaporating .
however , for an earth - like ocean , 10 years would not be enough to evaporate it all , so that the water reservoir would survive .
in case b ) , due to the excitation of the eccentricity of the orbits , the tidal heat flux of the inner planet always exceeds the runaway greenhouse limit . thus
, this hz planet would be too hot to host surface liquid water .
the middle planet ( in green ) is on the outer edge of the hz , and its tidal flux is relatively low ( in average it is about io s tidal flux ) . for case a ) , this planet could potentially have habitable conditions .
for case b ) , the planet spends some time around aphelion outside the insolation hz and it has a tidal heat flux a bit superior to io s in average . without tidal heating , this planet could be too cold to be able to sustain a potential liquid water reservoir , however taking into account tidal heating will improve the conditions for habitability .
one could imagine a more extreme case of a planet on an orbit completely outside the hz but heated up by tides sufficiently to be able to host surface liquid water .
when assessing habitability of planets in the hz of bds , one should investigate if tides are strong enough to drive a runaway greenhouse .
if the atmosphere receives an average flux ( @xmath18 ) lower than @xmath16 w / m@xmath11 , the planet can sustain a liquid water reservoir , but if it receives an average flux higher than @xmath16 w / m@xmath11 the planet will be too hot to be able to sustain a liquid water reservoir .
the planetary tide influences the rotation of the planet , which is an important parameter for climate studies . we take here the example of the planet kepler-186f , the first earth - sized planet in the hz of a star @xcite .
using the system parameters given in @xcite , we compute the tidal evolution of the system and focus our attention on the fifth planet . contrary to the four inner planets , which quickly reach a pseudo - synchronous state and low obliquities ( in @xmath19 myr ) , the evolution timescale of the rotation of kepler-186f is much longer .
figure [ fig3 ] shows the evolution of obliquity and rotation period for kepler-186f for different initial obliquities and rotation periods .
the initial values are of course not known but n - body simulations of terrestrial accretion tend to produce planets with fast initial spins and isotropically distributed obliquities @xcite .
[ fig3 ] figure [ fig3 ] shows that kepler-186f s obliquity increases for all but the slowest initial spin rate .
the obliquity increases for a few hundred megayears and this period is followed by a long , slow decay that lasts 2 - 3 gyr ( for the arbitrarily chosen range of initial spin rates and an earth - like tidal dissipation ) .
the equilibrium obliquity of the planet is very low , it does not feel the dynamical interactions of the four inner planet . given the estimated age of the system ( @xmath204 gyr according to * ? ? ?
* ) , kepler-186f could be in pseudo - synchronous rotation with a very small obliquity ( see fig .
[ fig3 ] ) .
but if the system is a bit younger ( 1 gyr ) , or if the planet dissipates less than what was assumed , kepler-186f could have a faster rotation and a very high obliquity ( @xmath21 degrees ) .
this would impact the climate of the planet : a non - negligible obliquity causes seasonal effects while a negligible obliquity causes a low insolation at the poles ; the rotation has an impact on the heat transport in the atmosphere and if it is sufficiently close to the synchronous rotation , the night side could a cold trap .
it is therefore necessary to consider all possible rotation states when assessing the climate and habitability of this planet .
the climate of a planet depends on many parameters .
it depends on the atmospheric pressure , temperature and composition but also on astronomic quantities : orbital distance ( insolation ) , eccentricity of the orbit ( insolation , seasons ) , rotation of the planet ( cold trap , heat redistribution ) , obliquity of the planet ( cold trap , seasons ) , tidal heat flux ( acting as an increase in insolation ) .
the stellar tide influences the orbital distance and the eccentricity of the planet while the planetary tide influences all these quantities .
there is therefore a very tight link between the dynamics of tidal evolution and the climate of a planet .
|
our knowledge of planets orbital dynamics , which was based on solar system studies , has been challenged by the diversity of exoplanetary systems . around cool and ultra cool dwarfs
, the influence of tides on the orbital and spin evolution of planets can strongly affect their climate and their capacity to host surface liquid water .
we illustrate the role of tides and dynamics with the extreme case of planets orbiting around brown dwarfs . in multiple planet systems ,
the eccentricity is excited by planet - planet interactions .
planets are therefore heated up from the inside by the tidally - induced friction .
this process can heat a habitable zone planet to such a level that surface liquid water can not exist .
we also talk about the newly discovered potentially habitable earth - sized planet kepler-186f . given the poorly estimated age of the system
, the planet could still be evolving towards synchronization and have a high obliquity or be pseudo - synchronized with a zero obliquity .
these two configurations would have a different effect on the climate of this planet .
planets and satellites : atmospheres , planets and satellites : dynamical evolution and stability , planets and satellites : individual : kepler-186f
|
since the pioneering paper by kosterlitz and thouless @xcite , the @xmath0 model is considered the classical example of discrete spin system exhibiting phase transitions mediated by the formation and the interaction of topological singularities . even at zero temperature
the model presents interesting features : depending on the energy regime , the geometry of the ground states is very rich , going from uniform to disordered states , all the way through isolated vortex singularities and clustered dipoles .
both topology and energy concentration take place at different length scales , thus making the analysis fascinating and popular also in the mathematical community .
in this paper we focus on the variational analysis of a two dimensional modified @xmath0 model at zero temperature , describing the formation of fractional vortices and string defects .
our main motivation comes from observing that this kind of singularities characterizes several discrete systems in physics and materials science such as disclinations and string defects in liquid crystals @xcite , fractional vortices and domain walls in micromagnetics @xcite , partial dislocations and stacking faults in crystal plasticity @xcite .
we do not focus on the specific details of any of such models , rather we aim at providing a simple variational model highlighting some of their relevant common features . given a bounded open set @xmath3 ,
the classical @xmath0 energy on the lattice @xmath4 is given by @xmath5 where the sum is taken over pairs of nearest neighbors @xmath6 ; i.e. , @xmath7 with @xmath8 , and @xmath9 is the spin field . by writing @xmath10 ,
the @xmath0 energy can be rewritten - up to constants - as @xmath11 where @xmath12 .
the rigorous upscaling as @xmath13 of the @xmath14 functional - and more in general of discrete spin systems governed by @xmath15-periodic potentials @xmath16 - has been recently obtained in @xcite in terms of @xmath2-convergence .
loosely speaking , configurations with energy of order @xmath17 exhibit a finite number ( controlled by the pre - factor @xmath18 ) of vortex - like singularities . around each singularity
the order parameter @xmath19 looks like a fixed rotation of the map @xmath20 where @xmath21 is the degree of the singularity .
there the energy concentrates and blows up as @xmath22 .
this analysis is then refined exploiting the next lower order term in the energy expansion .
indeed , after removing the logarithmic leading order contribution , a finite interaction energy , referred to as _ renormalized energy _ , remains .
the renormalized energy depends on the positions and the degrees of the vortices , and it is considered the main driving force responsible for their dynamics @xcite . the strategy adopted to analyze the @xmath0 model exploits methods and tools from the earlier analysis developed for the continuous ginzburg - landau functionals @xcite .
moreover , these techniques have been successfully used to understand in terms of @xmath2-convergence the well - known analogy between vortices in the @xmath0 model and in superconductivity , and screw dislocations in anti - plane linearized elasticity @xcite . to some extent
, this analogy extends _ mutatis mutandis _ to many other topological defects in liquid crystals , micromagnetics and crystal plasticity .
let us introduce and motivate our model using the language of planar uniaxial nematics .
in the oseen - frank and ericksen picture , nematics are described by a vector field @xmath23 , the _ director _ , that ( in two dimensions ) takes values in @xmath24 .
the molecules are supposed to interact via an elastic potential depending only on their orientation @xmath25 .
if the nematic molecules present _ head - to - tail symmetry _ , the interaction potential is @xmath26-periodic . therefore there is a clear identification between a @xmath27-dimensional nematic lattice model and the @xmath0 spin model .
doubling the phase variable @xmath25 of the director , the interaction potential becomes @xmath15-periodic so that the nematic energy can be written as in with @xmath28 .
this corresponds to the lebwohl and lasher functional ( @xcite ) recently studied by @xmath2-convergence in @xcite . since in the classical @xmath0 model
integer vortices appear , as result of the identification @xmath28 , for liquid crystals the relevant topological singularities are half vortices , namely the disclinations .
the presence of half vortices is related to orientability issues of the director , that can be better described and visualized in a continuous framework : in a configuration consisting of two half vortices , the director field can not be smooth everywhere . indeed , it has antipodal orientations on the two sides of a line joining the half vortices . according to @xcite
, we call this kind of discontinuity lines _ string defects _ ( see figure [ fig : dipoles ] )
. a first rigorous study of orientability issues related to the presence of half vortices can be found in @xcite , while the right space to describe discontinuous directors has been very recently identified in @xcite as a suitable subspace of @xmath29 , the space of _ special functions with bounded variation_. in the left picture and @xmath30 in right one ) .
the shaded regions highlight necessary antipodal spins forming the string defects . ]
( 0,0 ) in this paper we provide a quantitative analysis of such orientability issues , by introducing energy functionals which describe both half , and in fact fractional vortices , and string defects . to this end
, we propose a modified @xmath0 model , that draws back to @xcite .
we consider @xmath15-periodic potentials , acting on nearest neighbors , and having @xmath31 wells in @xmath32 ; in one well the potential is zero , while in the remaining @xmath33 it is positive but vanishing as @xmath34 goes to zero . in this way
, we model systems with one symmetry , and @xmath33 vanishing asymmetries . the case @xmath35 provides a simple toy model for liquid crystals with small head - to - tail asymmetry , like those made of non centrosymmetric or chiral molecules @xcite . more specifically , we consider potentials @xmath36 with @xmath33 wells of order @xmath34 ( see figure [ pot ] )
. having @xmath37 wells in the potential produces fractional vortices of degree @xmath38 , while the additional energy due to the weighted @xmath33 wells yields , together with the renormalized energy , a new term in the limit , depending on the length of the string defects .
the specific @xmath34-scaling of the wells makes finite the string defect energy , and hence of the same order of the renormalized energy .
other scalings could be considered , leading to different limit theories .
let us describe our model and results in more details .
given a discrete director field configuration @xmath39 , we introduce the modified @xmath0 energy functional @xmath40 we show that configurations @xmath39 with @xmath41 are consistent with the formation of @xmath42 fractional vortices at @xmath43 having degrees @xmath38 .
moreover , up to a subsequence , @xmath44 strongly converges in @xmath45 to some unitary field @xmath46 belonging to @xmath47 with @xmath48 . indeed in theorem [ mainthm ]
we prove that the functionals @xmath49 @xmath2-converge to the limit energy @xmath50 given by @xmath51 here @xmath52 represents the renormalized energy ; it depends on the positions and signs of the fractional vortices , but also on the excess energy due to unneeded oscillations of @xmath46 ; indeed , minimizing @xmath52 with respect to all director fields @xmath46 compatible with the given configuration of fractional vortices gives back the classical renormalized energy for vortices in superconductivity @xcite .
the constant @xmath53 is a core energy stored around each singularity , and has memory of the discrete lattice structure .
the third addendum is the anisotropic length of the string , reminiscent of the lattice symmetries .
we note that the analysis by @xmath2-convergence of discrete functionals leading to such anisotropic interfacial energies is a well established research field @xcite . on the other hand , limit energies as in
may appear also as @xmath2-limits of continuous ginzburg - landau type functionals .
actually , in @xcite the authors propose and study a variant of the standard ginzburg - landau energy , that could be seen as a continuous counterpart of our modified @xmath0 model .
we remark that our @xmath2-convergence result can be considered the first step towards the study of the joint dynamics of strings and vortices , as well as partial dislocations and stacking faults .
we expect that the competition between the surface and the renormalized energies generates non - trivial metastable configurations . as a matter of fact , neglecting boundary effects , two equally charged vortices - in the classical @xmath0 model - repel each other with a force proportional to the inverse of their distance , so that they are never in equilibrium .
on the other hand , since in the generalized @xmath0 model the surface term diverges with the distance between the vortices , there exists a critical length at which the two forces balance and equilibrium is reached . in this respect
, it seems interesting to investigate the dynamics of strings and fractional vortices driven by the limit energy functional @xmath54 , through both rigorous mathematical analysis and numerical simulations .
here we list some preliminaries on @xmath55 and @xmath29 functions that will be useful in the following .
we begin by recalling some standard notations .
let @xmath56 be open bounded with lipschitz boundary . as customary , @xmath57 ( resp .
@xmath58 ) denotes the set of functions of bounded variation ( resp .
special functions of bounded variation ) defined on @xmath59 and taking values in @xmath60 . moreover , we set @xmath61 and @xmath62 . given any set @xmath63 , the classes of functions @xmath64 and
@xmath65 are defined in the obvious way .
we refer to the book @xcite for the definition and the main properties of @xmath55 and @xmath29 functions . here
we recall that the distributional gradient @xmath66 of a function @xmath67 can be decomposed as : @xmath68 where @xmath69 is the approximate gradient of @xmath70 , @xmath71 is the jump set of @xmath70 , @xmath72 is a unit normal to @xmath71 and @xmath73 are the approximate trace values of @xmath70 on @xmath71 . for @xmath74 , @xmath75 denotes the subspace of @xmath58 defined by @xmath76 moreover , @xmath77 denotes the class of functions belonging to @xmath78 for all @xmath79 open .
we recall three results about compactness , lower semicontinuity and density properties of @xmath80 and @xmath81 functions .
[ sbvcomp ] let @xmath82 for some @xmath74 .
assume that there exists @xmath83 such that @xmath84 then , there exists @xmath85 such that , up to a subsequence , @xmath86 for any open set @xmath87 . in the following ,
we say that a sequence @xmath88 weakly converges in @xmath75 to a function @xmath89 , and we write that @xmath90 in @xmath75 , if @xmath91 satisfy and @xmath92 in @xmath93 .
moreover , we say that @xmath94 weakly converges in @xmath95 to a function @xmath96 , and we write that @xmath90 in @xmath95 , if @xmath90 in @xmath78 for any open set @xmath79 .
[ lscsbv ] let @xmath97 be compact and let @xmath98 .
let @xmath99 be a positive , symmetric function satisfying @xmath100 and let @xmath101 be an even , convex and positively 1-homogeneous function satisfying @xmath102 for all @xmath103 .
if @xmath104 and @xmath90 in @xmath77 for some @xmath105 , then @xmath106 finally , we state the density result proved in ( * ? ?
* theorem 2.1 and corollary 2.4 ) .
first we recall that a polyhedral set in @xmath107 is ( up to a @xmath108 negligible set ) a finite union of segments contained in @xmath107 .
[ polyapprox ] let @xmath109 be finite and let @xmath110 .
then there exists a sequence @xmath111 such that @xmath112 is polyhedral , @xmath113 in @xmath114 for any @xmath105 , and @xmath115 for any continuous function @xmath116 satisfying @xmath117 for all @xmath118 .
in particular , @xmath119 .
let @xmath31 be a fixed natural number , and let @xmath120 be an open bounded set with lipschitz continuous boundary , representing the domain of definition of the relevant fields in the models we deal with .
additionally , to simplify matter , we assume that @xmath107 is simply connected .
this is convenient in our constructions where we make use of poincar lemma ( as in proof of lemma [ bulklemma ] ) .
nevertheless , this assumption can be removed with some slight additional effort in our proofs , by introducing a finite number of `` cuts '' @xmath121 in @xmath107 , such that @xmath122 is simply connected . for every @xmath123 , we set @xmath124 where @xmath125 ^ 2 $ ] is the unit square
. moreover we set @xmath126 , and @xmath127\subset\omega_{\varepsilon}\right\}$ ] , where @xmath128 $ ] denotes the ( closed ) segment joining @xmath129 and @xmath130 .
these objects represent the reference lattice and the class of nearest neighbors , respectively .
the cells contained in @xmath131 are labeled by the set of indices @xmath132 .
finally , we define the discrete boundary of @xmath107 as @xmath133 in the following , we will extend the use of such notations to any given subset @xmath59 of @xmath134 . here
we introduce the classes of discrete functions on @xmath135 and a notion of discrete topological singularity .
we first set @xmath136 and we introduce the class of admissible fields from @xmath135 to the set @xmath24 of unit vectors in @xmath134 @xmath137 for any @xmath138 , @xmath139 , and for any @xmath140 we set @xmath141 following the formalism in @xcite , we introduce a notion of discrete vorticity for any given function @xmath142 . to this purpose ,
let @xmath143 be defined as follows @xmath144 with the convention that , if the argmin is not unique , then @xmath145 is the smallest one .
let @xmath142 be fixed .
for every @xmath140 , we define the signed distance of the discrete gradient of @xmath146 from @xmath147 , as @xmath148 where @xmath149 means that @xmath150 for @xmath151 .
notice that by definition @xmath152 for all @xmath153 . for every @xmath154 we define the discrete vorticity of the cell @xmath155 as @xmath156 one can easily see that the vorticity @xmath157 takes values in @xmath158 .
stokes theorem in our discrete setting reads as follows .
let @xmath159 with @xmath160 bounded and simply connected .
set @xmath161 and let @xmath162 , with @xmath163 for any @xmath164 , and notice that @xmath165 .
then , for any @xmath142 , it holds @xmath166 we define the vorticity measure @xmath167 as follows @xmath168 let @xmath56 . for any @xmath169 with @xmath170 , @xmath171 and @xmath172 , the _ flat norm _ of @xmath173 is defined as @xmath174 whenever it will be convenient , we will declare the domain @xmath59 of the test functions @xmath175 in the definition of the flat norm by writing @xmath176 instead of @xmath177 .
we will denote by @xmath178 the flat convergence of @xmath179 to @xmath173 .
it is well - known that the flat norm of @xmath173 is related to integer @xmath180-currents @xmath181 with @xmath182 and having minimal mass ( for the theory and terminology of integer currents we refer the reader to @xcite ) .
more precisely , by ( * ? ? ?
* section 4.1.12 ) ( see also ( * ? ? ? * formula ( 6.1 ) ) ) there holds @xmath183 where @xmath184 is the set of functions @xmath175 with @xmath185 . since for every @xmath186 we have @xmath187 we deduce that @xmath188 here we will introduce a class of energy functionals defined on @xmath189 .
let @xmath190 be a continuous @xmath191-periodic function such that * @xmath192 if and only if @xmath193 ; * @xmath194 for any @xmath195 ; * @xmath196 as @xmath197 . and set @xmath198 . for any fixed @xmath123
, we consider the pairwise interaction potentials @xmath199 defined by @xmath200}(t).\ ] ] we stress that the analysis done in this paper might be extended with minor changes to a larger class of potentials such as the smooth ones considered in @xcite . for any @xmath201 , we define @xmath202 for the convenience of the reader , we now introduce two additional energy functionals , which will be useful in the proof of our main result . we define @xmath203 let @xmath204 and set @xmath205 , @xmath206 and @xmath207 . by ( ii ) , we have that @xmath208 in the following we will consider also the localized version of the energy @xmath209 , defined for any set @xmath210 by @xmath211 the localized versions of the energies @xmath212 and @xmath213 on a set @xmath210 are analogously defined and are denoted by @xmath214 and @xmath215 , respectively . here
we introduce the extensions of our discrete order parameters to the whole domain . in order to extend the discrete functions from @xmath135 to the whole @xmath216
, we will use two types of interpolations depending on whether the function is scalar or vector valued . to this purpose
, we consider the triangulation defined by the following sets : @xmath217 where , for any @xmath218 , @xmath219 denotes their convex envelope .
let @xmath220 . according to
, we define the piecewise affine interpolations of @xmath221 in an as follows : for any @xmath222 and for any @xmath223 , we set @xmath224 \!\!\!\ !
w(i)\!+\!\frac{\operatorname{d}w(i+{\varepsilon}e_2,i+{\varepsilon}e_1+{\varepsilon}e_2)}{{\varepsilon}}(x_1-i_1)\!+\!\frac{\operatorname{d}w(i , i+{\varepsilon}e_2)}{{\varepsilon}}(x_2-i_2),\,\textrm{if } x\in { t}_i^+ . \end{array}\right.\ ] ] one can easily check that , for any open set @xmath210 and for any @xmath220 , @xmath225 let @xmath226 .
we say that @xmath140 is a jump pair of @xmath25 , if @xmath227 and we denote by @xmath228 the set of jump pairs of @xmath25 .
furthermore we say that @xmath229 ( @xmath230 ) is a jump cell , if @xmath231 is a jump pair of @xmath25 for some bond @xmath232 we denote by @xmath233 the set of jump cells of @xmath25 .
now , recalling that @xmath234 , we set @xmath235 finally , recalling that @xmath236 , we set @xmath237 . with a little abuse of notations
we identify the functions @xmath238 and @xmath239 with @xmath240 functions defined on @xmath107 , just by extending them to @xmath241 in @xmath242 .
in this section we state our @xmath2-convergence result for the energies @xmath243 defined in . to this purpose , we precisely define its @xmath2-limit @xmath244 .
as mentioned in the introduction , the functional @xmath244 is given by the sum of three terms : the _ renormalized energy _
@xmath245 , representing the energy far from the limit singularities , the core energy @xmath53 , and the anisotropic surface term measuring the length of the string defects .
we start by defining the renormalized energy @xmath245 .
fix @xmath246 , we set @xmath247 where @xmath248 denotes the distributional jacobian of @xmath249 ( see for instance @xcite ) , defined by @xmath250 as shown in @xcite , if @xmath251 with @xmath252 , then the function @xmath253 is monotonically decreasing with respect to @xmath254 .
therefore , it is well defined the functional @xmath255 given by @xmath256 we introduce also the localized version of the energy @xmath245 . for any open @xmath257 with @xmath258 , we consider @xmath259 defined as in with @xmath107 replaced by @xmath260 , and set @xmath261 set @xmath262 where @xmath263 denotes the degree of @xmath19 on @xmath264 ( see for instance @xcite ) . as shown in @xcite , the renormalized energy @xmath265 in
is well defined for any @xmath266 .
moreover , by ( * ? ? ?
* remark 4.4 ) , if @xmath267 , then the dirichlet energy of @xmath19 is uniformly bounded on all diadic annuli around each @xmath268 . by using hlder inequality ( on each diadic annulus )
we obtain that @xmath269 .
in particular the class of functions in @xmath270 with @xmath267 coincides with the class of functions in @xmath271 with @xmath267 .
finally , we notice that if @xmath272 with @xmath273 , integration by parts easily yields @xmath274 in order to define the `` finite core energy '' @xmath53 , we consider an auxiliary minimization problem . given @xmath275 , we set @xmath276 where @xmath277 is defined in and @xmath278 denotes the angular polar coordinate for any @xmath279 . moreover , we set @xmath280 by ( * ? ? ?
* theorem 4.1 ) the above limit exists , is finite and does not depend on @xmath254 .
finally , for any @xmath246 , we define the domain of the @xmath2-limit as @xmath281 and for any @xmath282 we set @xmath283 where , for any @xmath284 , @xmath285 denotes the @xmath180-norm of @xmath286 .
we are now in a position to state our @xmath2-convergence result .
we recall that the assumptions on the interaction potentials @xmath287 are listed in subsection [ discreteenergies ] .
moreover , given @xmath288 we recall that @xmath205 and @xmath289 . in order to ease the notations , for any function @xmath290 , we will denote the fields @xmath291 ( introduced in subsection [ extdiscrfun ] ) by @xmath292 , respectively ; for the same reason , we will denote by @xmath293 the discrete vorticity measure @xmath294 .
[ mainthm ] let @xmath295 be fixed .
the following @xmath2-convergence result holds true .
\(i ) ( compactness ) let @xmath296 be a sequence satisfying @xmath297 for some @xmath298 . then , up to a subsequence , @xmath299 with @xmath170 , @xmath300 , @xmath301 , @xmath302 for @xmath303 and @xmath304 . moreover , if @xmath305 , then @xmath306 and there exists @xmath307 with @xmath308 such that , up to a further subsequence , @xmath309 in @xmath310 .
\(ii ) ( @xmath2-liminf inequality ) let @xmath311 and let @xmath312 be such that @xmath313 and @xmath314 in @xmath315 .
then , @xmath316 \(iii ) ( @xmath2-limsup inequality ) given @xmath317 , there exists @xmath312 such that @xmath313 , @xmath318 in @xmath319 and @xmath320 a similar @xmath2-convergence result for the functional @xmath321 has been proved in @xcite . for our purposes
it is convenient to present here its precise statement using our notations .
[ gammal2 ] let @xmath246 be fixed .
let @xmath322 be such that @xmath323 , for some @xmath298 .
then , up to a subsequence , @xmath324 with @xmath170 , @xmath171 , @xmath325 , @xmath302 for @xmath303 and @xmath326 ; moreover , there exists a constant @xmath327 such that for any @xmath328 and for any @xmath329 , there holds @xmath330 in particular , @xmath331 furthermore , if @xmath332 , then * @xmath306 for any @xmath328 ; * up to a further subsequence , @xmath333 in @xmath334 for some @xmath335 with @xmath336 ; * @xmath337 , for any open @xmath257 with @xmath258 . since in @xcite
it is not explicitly stated that @xmath338 , for the reader s convenience we provide here a short proof of this fact . as proven in @xcite , if @xmath339 , then @xmath340 now let @xmath341 , @xmath324 with @xmath342 and @xmath325 , @xmath302 for @xmath303 , and @xmath333 in @xmath334 for some @xmath335 .
let @xmath343 and let @xmath344 be analogously defined with @xmath19 replaced by @xmath345 .
then , @xmath346 , @xmath347 , and @xmath348 in @xmath349 .
let @xmath254 be as in and let @xmath350 $ ] be the piecewise affine function such that @xmath351 in @xmath352 $ ] , @xmath353 in @xmath354 and @xmath175 is affine in @xmath355 $ ] .
fix @xmath356 and set @xmath357 for any @xmath358 .
then , in view of , we have @xmath359 which combined with the fact that @xmath360 in @xmath361 , yields @xmath338 .
this section is devoted to the proof of theorem [ mainthm ] . in
what follows the letter @xmath18 will denote a constant in @xmath362 that may change from line to line ; if the constant @xmath18 will depend on some parameters @xmath363 we will write @xmath364 .
moreover , for any @xmath365 , for any @xmath210 open and for any @xmath366 , we set @xmath367 by and by , we immediately get @xmath368 therefore , by theorem [ gammal2 ] , we have that , up to a subsequence , @xmath369 for some @xmath173 with all the desired properties .
let us assume that @xmath370 . by theorem [ gammal2](a )
, we get that @xmath342 , and by theorem [ gammal2](b ) , up to passing to a further subsequence , @xmath333 in @xmath334 for some @xmath335 with @xmath338 .
now we prove that @xmath309 in @xmath310 , for some @xmath307 with @xmath371 , and hence , in particular , @xmath308 .
consider an increasing sequence @xmath372 of open smooth sets compactly contained in @xmath107 such that @xmath373 .
fix @xmath374 ( large enough ) and let @xmath365 ( small enough ) be such that the balls @xmath375 are pairwise disjoint and contained in @xmath376 .
we first prove that @xmath377 satisfies the upper bound in @xmath378 , i.e. , that @xmath379 for some constant @xmath380 .
let @xmath123 be small enough so that @xmath381 .
we preliminarly notice that @xmath382 moreover , ( recalling the notations introduced in subsection [ extdiscrfun ] ) , we set @xmath383 it is easy to check , that for any @xmath140 with @xmath384 , it holds @xmath385 for some @xmath386 . using that @xmath387 is piecewise constant in @xmath388 , and , we get @xmath389 where the last inequality is a consequence of and , and @xmath390 .
we now show that @xmath391 for some constant @xmath18 independent of @xmath392 , @xmath393 and @xmath34 .
first , it is easy to see that @xmath394 then , in order to prove , it is enough to prove that @xmath395 is uniformly bounded with respect to @xmath34 , @xmath392 and @xmath393 . by , and , we have @xmath396 since @xmath397 we have @xmath398 in @xmath399 . by lower semicontinuity , we obtain @xmath400 where @xmath401 .
this , together with yields @xmath402 and in turn .
therefore , by , and , the bound is satisfied and by theorem [ sbvcomp ] there exists a unitary field @xmath403 such that , up to a subsequence , @xmath404 in @xmath405 . by a standard diagonal argument , up to a further subsequence , @xmath318 in @xmath406 , for some @xmath407 .
moreover , by , @xmath408 is finite . in order to complete the proof it remains to show that @xmath409 . to this purpose
, we set @xmath410 . let @xmath230 with @xmath411 and let @xmath412 , with @xmath413 defined as in . by
, we have @xmath414 which combined with the fact that @xmath415 and the mean value theorem , yields @xmath416 using the same argument one can show that for any @xmath417 @xmath418 by integrating and using , we obtain @xmath419 which , sending @xmath13 , yields @xmath420 in @xmath421 .
now , since @xmath314 in @xmath315 , we clearly have that @xmath422 in @xmath315 , so that we conclude @xmath371 .
we can assume without loss of generality that holds and that @xmath423 converges ; then , by the compactness result @xmath424 of theorem [ mainthm ] , @xmath309 in @xmath425 for some @xmath307 .
set @xmath426 . since @xmath427 , @xmath428 with @xmath342 and @xmath325 for any @xmath328
let @xmath365 be such that the balls @xmath429 are pairwise disjoint and contained in @xmath107 .
as @xmath430 is rectifiable , it is contained , up to a @xmath108-negligible set , in a countably union @xmath431 of compact @xmath432-curves contained in @xmath433 .
for every @xmath434 , we define @xmath435 . then , for any given @xmath436 we define the @xmath437-tube around @xmath438 in @xmath439 ( see figure [ tubefig ] ) as @xmath440 moreover , we set @xmath441 . notice that for all @xmath123 and for @xmath437 small enough , @xmath442 has lipschitz continuous boundary .
it remains to prove the claim .
fix @xmath448 and let @xmath449 be defined by @xmath450 it is straightforward to check that @xmath451 is positive , symmetric and satisfies the triangular inequality .
let @xmath452 be a sequence of open sets with @xmath453 for all @xmath454 and such that @xmath455 . since @xmath456 on @xmath430 , we have that @xmath457 on @xmath430 , whence , recalling also theorem [ lscsbv ] , we get @xmath458 therefore , in order to get it is enough to prove that @xmath459 for some @xmath83 . to this purpose
, we set @xmath460{\varepsilon}\,\textrm { for some } \right.\\ & \left.(j , k ) \in { \omega}_{\varepsilon}^1 \text { with } \{j , k\ } \cap \partial ( i+{\varepsilon}{q } ) \neq \emptyset \right\}.\end{aligned}\ ] ] indeed , by the energy bounds and , there exists @xmath462 such that @xmath463 therefore , by the definition of @xmath464 there exists @xmath83 such that @xmath465{\varepsilon } ) ) \\ \leq\sharp \ { ( j , k)\in ( t_{{\varepsilon}}^{\mathrm{ng}})_{{\varepsilon}}^1 : \ , \operatorname{dist}({\varphi}_{\varepsilon}(k ) - { \varphi}_{\varepsilon}(j),\textstyle \frac{2\pi } { { \bf n}}{\mathbb{z } } ) >
\sqrt[3]{\varepsilon}\ } ( 1-\cos({{\bf n}}\sqrt[3]{\varepsilon } ) ) \\
\leq xy_{\varepsilon}(v_{\varepsilon } , t_{{\varepsilon}}^{\mathrm{ng } } ) \leq c_\rho,\end{gathered}\ ] ] which by taylor expansion yields @xmath466 and eventually @xmath467 then follows by noticing that @xmath451 is uniformly bounded by @xmath180 .
consider the map @xmath468 defined by @xmath469 and @xmath470 for all @xmath471 .
moreover , for all @xmath472 , let @xmath473 .
now , set @xmath474 we show that for any @xmath475 and for any @xmath476 $ ] , there holds @xmath477 , the proof in the case @xmath478 being analogous .
notice that it is trivial whenever @xmath155 and @xmath479 belong to the family of jump cells @xmath480 ( see figure [ edge_vs_pair ] ) , since in this case for any @xmath481 $ ] @xmath482 analogously , is trivial whenever both @xmath155 and @xmath479 are not jump cells , since in this case @xmath483 and ( recall @xmath484 ) @xmath485{{\varepsilon}}$ ] .
now , without loss of generality we assume that ( see figure [ edge_vs_pair ] ) @xmath155 is not a jump cell while @xmath479 is a jump cell ( the proof being fully analogous in the case that @xmath479 is not a jump cell while @xmath155 is a jump cell ) .
furthermore , we prove only the second inequality in since the first one follows similarly .
let @xmath486 $ ] be such that @xmath487 . then , using that @xmath488 and @xmath489 , we have @xmath490{\varepsilon}}{2 } \le |u_{\varepsilon}(i)-u_{\varepsilon}(i-{\varepsilon}e_2)|+\sqrt[3]{\varepsilon}.\end{gathered}\ ] ] in conclusion we have proven that holds true .
notice now that if @xmath475 , then @xmath491 , and hence @xmath492 satisfies either @xmath493{{\varepsilon}}\ ] ] or @xmath494{{\varepsilon}},\ ] ] for some @xmath495 .
if @xmath496 satisfies , then , by , @xmath497}|\hat u^+_{\varepsilon}(x)-\hat u^-_{\varepsilon}(x)|\le |u_{\varepsilon}(\jmath)-u_{\varepsilon}(\imath)|+\sqrt[3]{{\varepsilon}}\\ = 2|\sin\textstyle \frac{{\varphi}_{\varepsilon}(\jmath)-{\varphi}_{\varepsilon}(\imath)}{2}|+\sqrt[3]{{\varepsilon}}\le 2\sqrt[3]{{\varepsilon}}.\ ] ] on the other hand , if @xmath496 satisfies for some @xmath498 , then , again by , and using taylor expansion , for any @xmath499 $ ] and for @xmath34 sufficiently small , we have @xmath500{{\varepsilon}}\le 2|\sin\frac{{\varphi}_{\varepsilon}(\jmath)-{\varphi}_{\varepsilon}(\imath)}{2}|-\sqrt[3]{{\varepsilon}}\\ = |u_{\varepsilon}(\jmath)-u_{\varepsilon}(\imath)|-\sqrt[3]{{\varepsilon}}\le |\hat u^+_{\varepsilon}(x)- \hat u_{\varepsilon}^-(x)|\le |u_{\varepsilon}(\jmath)-u_{\varepsilon}(\imath)|+\sqrt[3]{{\varepsilon}}\\ \le2|\textstyle \sin\frac{l'\pi}{{{\bf n}}}|+2\sqrt[3]{{\varepsilon}}.\end{gathered}\ ] ] it follows that , if @xmath34 is sufficiently small , the jump bonds of @xmath387 in @xmath501 , and , a fortiori , the jump bonds of @xmath387 in @xmath502 lie in one of the following sets @xmath503}|\hat u^+_{\varepsilon}(x)-\hat u^-_{\varepsilon}(x)|\le 2\sqrt[3]{{\varepsilon}}\},\\ & & i_{\delta,\rho,{\varepsilon}}^{b}:=\{(i , j)\in(\tilde t^n_{\delta - \sqrt{2}{\varepsilon},\rho,{\varepsilon}})_{\varepsilon}^1\,:\,\left|\textstyle |\hat u^+_{\varepsilon}(x)-\hat u^-_{\varepsilon}(x)|-2\sin\frac{l\pi}{{{\bf n}}}\right|\le 2\sqrt[3]{{\varepsilon}}\\ & & \phantom{i_{\delta,\rho,{\varepsilon}}^{bj}:=}\textrm { for any } x\in[i , j ] , \textrm { for some } l=1,\ldots , { { { \bf n } } } -1 \}.\end{aligned}\ ] ] set @xmath504 , \qquad e^b:= \bigcup_{(i , j)\in i_{\delta,\rho,{\varepsilon}}^{b}}[i , j].\ ] ] by the very definition of @xmath451 and by the uniform bound of @xmath505 in , for @xmath34 small enough we have @xmath506{{\varepsilon}})\mathcal h^1(s_{\hat u_{\varepsilon}})\le c ( t + \sqrt[3]{{\varepsilon}}).\ ] ] in order to conclude the proof of , we first show that @xmath507 to this purpose , we set @xmath508{{\varepsilon}}\\ \textrm { for some } l'\in{\mathbb{z}}\setminus{{\bf n}}{\mathbb{z}}\},\end{gathered}\ ] ] notice that @xmath509 .
now we show that , for @xmath34 small enough , @xmath510 .
let @xmath511 and let @xmath512 be such that , for all @xmath476 $ ] we have @xmath513{\varepsilon}\le |u_{\varepsilon}(\jmath)-u_{\varepsilon}(\imath)|\le 2\sin\frac{l\pi}{{{\bf n } } } + 3\sqrt[3]{\varepsilon}.\ ] ] on the other hand , since @xmath514 , there exists @xmath515 such that @xmath516{\varepsilon}.\ ] ] therefore , by and and taylor expansion , it immediately follows that , for @xmath34 small enough , @xmath517 and hence @xmath518 .
we have concluded the proof that @xmath510 . by , and , we deduce @xmath523{{\varepsilon } } ) \\ \le\liminf_{{\varepsilon}\to 0 } { \mathcal{f}}_{\varepsilon}^{{({{\bf n}})}}({\varphi}_{\varepsilon},{t^n_{\delta,\rho,{\varepsilon } } } ) + ct,\end{gathered}\ ] ] which concludes the proof of the inequality in and of the @xmath2-liminf inequality .
we start this subsection by proving two lemmas that will be useful in the proof of the @xmath2-limsup inequality . the first lemma concerns with the first - order core energy induced by a singularity with degree @xmath524 .
we recall that @xmath525 denotes the angular polar coordinate in @xmath526 . to this purpose fix @xmath538 . for any @xmath123 ,
let @xmath539 be a solution of the minimization problem in and set @xmath540 . by , ( for @xmath541 )
we have @xmath542 .
let @xmath543 be the extension of @xmath539 defined by @xmath544 where @xmath545 is the angular polar coordinate .
notice that @xmath546 so that @xmath547 we show that there exists @xmath548 such that @xmath549 notice that , by standard interpolation estimates @xcite , @xmath550 therefore , for sufficiently small @xmath34 , we have @xmath551 by theorem [ gammal2 ] , we have that , up to a subsequence , @xmath552 where either @xmath553 or @xmath554 for some @xmath555 and @xmath556
. by there exist integer 1-currents @xmath561 , whose support is a finite union of segments , with @xmath562 and @xmath563 let moreover @xmath564 be an integer @xmath180-current with @xmath565 and @xmath566 ( such a current can be constructed , for instance , by considering an oriented segment @xmath567 joining @xmath568 with a point in @xmath569 ) .
set @xmath570 ; by construction @xmath571 and , by , @xmath572 set @xmath573 and @xmath574 .
let @xmath575 denote the connected components of @xmath576 .
let @xmath577 be such that @xmath160 is simply connected and @xmath578 .
then , it is easy to see that @xmath579 . by
, it follows that @xmath580 where @xmath161 and @xmath162 , with @xmath163 for any @xmath164 .
therefore , it is well - defined the function @xmath581 constructed as follows : for any @xmath582 fix @xmath583 and for any @xmath584 , set @xmath585 where @xmath586 is such that @xmath587 for @xmath588 . :
the line segments denote the support of @xmath561 .
all cells having positive ( resp . , negative )
vorticity @xmath589 are colored red ( resp . ,
blue ) . it can be easily seen that any closed path in @xmath590 is surrounding the same amount red and blue cells , implying that the definition of @xmath591 in is well - posed.,scaledwidth=40.0% ] indeed , by , one can easily check that @xmath592 does not depend on the path @xmath593 ( see also figure [ corefig ] ) .
moreover , since for any @xmath594 there exists @xmath582 such that @xmath595 , there holds @xmath596 furthermore , by the very definition of @xmath597 , it immediately follows that @xmath598 for any @xmath599 , we set @xmath600 by , @xmath601 is a competitor for the minimum problem . in order to prove ,
it is enough to show that @xmath602 with @xmath537 . in order to get
we first notice that by , for any @xmath594 , there holds @xmath603 whence , using and the very definition of @xmath604 in , we obtain @xmath605 moreover , again by the definition of @xmath604 , we have @xmath606 set @xmath607 , by construction and by , we have that @xmath608 in the end , by , and , we get @xmath609 which proves and concludes the proof of the lemma .
now we pass to a lemma which allows to deal with the far - field energy , i.e. , with the energy outside suitable balls centered in the limit singularities . to this purpose , for any @xmath427 , we denote , as usual , @xmath613 by @xmath614 ( see and ) . moreover , we recall that for any @xmath210 open and for any @xmath365 , @xmath615 is defined according to .
[ bulklemma ] let @xmath427 and let @xmath616 be pairwise disjoint segments such that @xmath617 joins @xmath268 with @xmath618 and @xmath619 for any @xmath328 .
fix @xmath365 , and set @xmath620 .
then , there exist @xmath621 with @xmath622 = \pm\frac{2\pi}{{{\bf n}}}$ ] on each @xmath617 and @xmath623 , such that @xmath624 is a lifting of @xmath46 in @xmath439 , i.e. , @xmath625 .
in particular , for any @xmath328 @xmath626+[w]){\mathop{\hbox{\vrule height 7pt width .5pt
depth 0pt \vrule height .5pt
width 6pt depth 0pt}}\nolimits}\gamma_i\in 2\pi{\mathbb{z}}.\ ] ] _ step 1 : construction of @xmath221 and @xmath146 .
_ let @xmath640 be a lifting of @xmath46 .
such a lifting always exists in virtue of ( * ? ? ?
* remark 4 ) . by
* chapter 1 ) , we have that @xmath641 is a conservative vector field on @xmath642 , and hence there exists a function @xmath621 such that @xmath643 . moreover , using that @xmath308 one can easily show that for any @xmath328 @xmath644{\mathop{\hbox{\vrule height 7pt width .5pt
depth 0pt \vrule height .5pt
width 6pt depth 0pt}}\nolimits}\gamma_i=\pm\frac{2\pi}{{{\bf n}}}d_i,\ ] ] the sign depending on the orientation of the normal to @xmath617 . since @xmath645 in @xmath642 , there exists a caccioppoli partition @xmath646 of @xmath439 subordinated to @xmath647 , such that @xmath648 for some @xmath649 . by and by the fact that @xmath650 , it follows that there exists a constant @xmath651 such that @xmath652 for any @xmath653 .
up to replacing @xmath221 with @xmath654 , we can always assume that @xmath655 .
for any @xmath656 we set @xmath657 set moreover @xmath658 . by construction
@xmath659 moreover , again by construction , @xmath660 ; therefore , @xmath624 is a lifting of @xmath46 , and is satisfied .
_ step 2 : approximation of @xmath221 .
_ by , the function @xmath661 belongs to @xmath662 .
therefore , by @xcite , there exists @xmath663 such that @xmath664 in @xmath665 .
it is well known ( see for instance @xcite ) that for @xmath393 sufficiently large @xmath666 for any @xmath328 . since @xmath107 is simply connected , also @xmath667 is and hence there exists @xmath668 such that @xmath669 . recalling that @xmath670
, we have that @xmath671 ( where we have identified @xmath134 with @xmath672 ) and hence @xmath673 this fact , combined with poincar inequality , yields @xmath674 for any @xmath374 set @xmath675 then , @xmath627 , @xmath628=[w]$ ] and ( i ) is satisfied .
_ step 3 : approximation of @xmath146_. for every @xmath328 , we denote by @xmath676 the intersection point of @xmath617 with @xmath618 and by @xmath677 a ( fixed ) unit normal vector to @xmath617 .
fix @xmath678 . for any @xmath679 and for any @xmath328
, we set @xmath680 moreover , for any @xmath681 we set @xmath682 we notice that for @xmath681 small enough , @xmath683 are pairwise disjoint and contained in @xmath439 and @xmath684 is lipschitz continuous .
then , by applying theorem [ polyapprox ] with @xmath685 , @xmath686 , and @xmath687 , we have that there exists a sequence @xmath688 such that , for any @xmath374 , @xmath689 is polyhedral and the following properties are satisfied @xmath690 @xmath691 and @xmath692 for all continuous @xmath693 satisfying @xmath694 .
we now extend @xmath695 from @xmath696 to @xmath439 by reflection along the segments @xmath697 and @xmath698 in the following manner ( see also figure [ reflection ] ) .
it is easy to see that @xmath705 and that @xmath706 is polyhedral .
let @xmath707 ( resp . ,
@xmath708 ) denote the trace of @xmath709 on @xmath710 ( resp . , @xmath711 ) .
by theorem [ polyapprox ] @xmath712 and hence , ( see for instance ( * ? ? ? * theorem 2.11 ) ) , we obtain @xmath713 moreover , we also have that for any @xmath328 @xmath714 for any @xmath328 , we set @xmath727 and we denote by @xmath728 ( resp . , @xmath729 ) the interior of @xmath730 ( resp . , @xmath731 ) .
fix @xmath328 .
again by construction and by , we have that @xmath732 with @xmath733 .
moreover , since by construction @xmath734=0 $ ] on @xmath735 , we have @xmath736 with @xmath737 . by and , and by triangular inequality
, it follows that @xmath738 then follows by summing and .
finally we prove .
fix @xmath328 . by and
, we have @xmath739 then , by dominated convergence theorem , we obtain @xmath740 with @xmath741 .
it follows that @xmath742 by summing over @xmath328 , we obtain .
this concludes the proof of .
_ step 5 : conclusion _ using and , we show that ( ii ) and ( iii ) hold true . by a standard diagonal argument in @xmath393 , @xmath437 and @xmath175 ,
there exists a sequence of functions @xmath743 with polyhedral jump set , such that @xmath631 satisfies ( iii ) .
moreover ( ii ) follows by theorem [ sbvcomp ] , up to extracting a further subsequence .
we notice that in lemma [ bulklemma](iii ) the assumption that @xmath719 is bounded can be replaced by requiring that @xmath744 , for some continuous functions @xmath745 and @xmath746 .
in fact , the former assumption is used only in the proof of .
now we are in a position to prove the @xmath2-limsup inequality .
we recall that for any @xmath747 , @xmath748 , with @xmath342 and @xmath325 . to ease the notations , for any @xmath328 , we set @xmath749 , where @xmath545 is the angular polar coordinate ( with respect to the origin ) . without loss of generality
we can assume that @xmath750 and hence there exists a constant @xmath83 such that for any @xmath538 @xmath751 fix @xmath538 and let @xmath752 denote the annulus @xmath753 for any @xmath754 and for any @xmath328 .
arguing as in the proof of ( * ? ? ?
* theorem 4.5 ) ( see also @xcite ) , one can show that for any @xmath754 there exists a unitary vector @xmath755 such that @xmath756 for any lifting @xmath25 of @xmath46 with @xmath757 ( a lifting like that exists by ( * ? ? ?
* remark 4 ) ) . fix @xmath754 and let @xmath758 and @xmath759 be the sequences constructed in lemma [ bulklemma ] for @xmath760 . for any @xmath374 , set @xmath761 .
let moreover @xmath762 be the lifting of @xmath46 constructed in lemma [ bulklemma ] with @xmath760 . by and by lemma [ bulklemma](i ) , using also that @xmath763 for a.e .
@xmath764 , it follows that @xmath765 with @xmath766 . by construction and recalling that @xmath767=[w_k]=\frac{2\pi}{{{\bf n}}}$ ] , we have that @xmath768 is a conservative field in @xmath769 , and so for any @xmath770 there exists a function @xmath771 with null average such that @xmath772 by poincar inequality and by , we have @xmath773 for some @xmath83 depending only on @xmath254 ( and not on @xmath774 ) .
let @xmath775;[0,1]\right)$ ] be such that @xmath776 for @xmath777 $ ] and @xmath778 for @xmath779 $ ] , and set @xmath780 set @xmath781 . by and , it easily follows that @xmath782 we define @xmath783 since @xmath784 , we have that for any @xmath328 @xmath785 whence @xmath786 moreover , for any @xmath328 , let @xmath787 be a solution of the minimum problem @xmath788 with @xmath789 on @xmath790 . by a standard density argument
we can always assume that @xmath791 , so we can set @xmath792 set @xmath793 and let @xmath794 denote the piecewise affine interpolation of @xmath795 constructed according with . in order to conclude the proof it is enough to show @xmath796 indeed , by , and theorem [ sbvcomp ] , using a standard diagonal argument
, there exists a sequence @xmath797 satisfying and with @xmath318 in @xmath798 .
we now prove and . to this purpose set @xmath799 by standard interpolation estimates ( see for instance @xcite ) , we have @xmath800 which combined with lemma [ bulklemma ] yields @xmath801 moreover , by lemma [ bulklemma ] , ( for @xmath393 sufficiently large ) @xmath802 since @xmath803 , by @xmath804 therefore , immediately follows by and .
now we pass to the proof of .
we set @xmath805\cap\cup_{i=1}^m \partial c_{i , k}\neq\emptyset\ } , \ ] ] and @xmath806 we notice that @xmath807 we investigate the convergence of the terms @xmath808 separately . by construction and by standard interpolation estimates
, we have @xmath809 by and again by standard interpolation estimates @xmath810 by construction , @xmath811 by standard interpolation estimates , , and lemma [ bulklemma ] , we have @xmath812 now we prove that @xmath813 by construction , for any @xmath814 we have @xmath815 where the last inequality follows by the very definition of @xmath816 in ( see also ) .
therefore , by the very definition of @xmath604 ( see ) , for @xmath34 sufficiently small there holds @xmath817 let @xmath818 be defined by @xmath819 notice that given any segment @xmath820 with normal @xmath821 , we have that @xmath34 times the number of the bonds in @xmath822 intersecting @xmath823 converges , as @xmath13 , to @xmath824 .
this fact , together with , yields @xmath825 therefore , by using lemma [ bulklemma ] ( iii ) and the fact that @xmath826=[{\varphi}_{k , h}]$ ] ( see and ) , we obtain @xmath827 which is exactly
. * acknowledgements . * the work of rb , mc , ldl was supported by the dfg collaborative research center trr 109 , `` discretization in geometry and dynamics '' .
part of this work was carried out while mp was visiting tum thanks to the visiting professors program funded by the bavarian state ministry for science , research and the arts .
r. alicandro , l. de luca , a. garroni , m. ponsiglione : metastability and dynamics of discrete topological singularities in two dimensions : a @xmath2-convergence approach , _ arch .
rational mech .
* 214 * ( 2014 ) , no .
1 , 269330 .
de luca , l. : @xmath2-convergence analysis for discrete topological singularities : the anisotropic triangular lattice and the long range interaction energy , _ asymptot .
* 96 * ( 2016 ) , no .
|
we propose and analyze a generalized two dimensional @xmath0 model , whose interaction potential has @xmath1 weighted wells , describing corresponding symmetries of the system . as the lattice spacing vanishes , we derive by @xmath2-convergence the discrete - to - continuum limit of this model . in the energy regime
we deal with , the asymptotic ground states exhibit fractional vortices , connected by string defects .
the @xmath2-limit takes into account both contributions , through a renormalized energy , depending on the configuration of fractional vortices , and a surface energy , proportional to the length of the strings .
our model describes in a simple way several topological singularities arising in physics and materials science . among them ,
disclinations and string defects in liquid crystals , fractional vortices and domain walls in micromagnetics , partial dislocations and stacking faults in crystal plasticity . * keywords : * @xmath0 spin systems , ginzburg - landau , liquid crystals , dislocations , calculus of variations , @xmath2-convergence + * mathematical classification : * 82d30 , 82b20 , 49j45
|
leclerc and zelevinsky @xcite , in their study of quasicommuting families of quantum minors , introduced the following notion of weakly separated sets .
let @xmath0 and @xmath1 be two subsets of @xmath2:=\ { 1,2 , \ldots , n \}$ ] .
leclerc and zelevinsky @xcite , defined @xmath0 and @xmath1 to be _ weakly separated _ if either 1 .
@xmath3 and @xmath4 can be partitioned as @xmath5 such that @xmath6 , or 2 .
@xmath7 and @xmath8 can be partitioned as @xmath9 such that @xmath10 , where @xmath11 indicates that every element of @xmath12 is less than every element of @xmath13 .
leclerc and zelevinsky proved that , for any collection @xmath14 of pairwise weakly separated subsets of @xmath2 $ ] , one has @xmath15 .
moreover , they made the following _ purity conjecture .
_ [ conj : lz ] @xcite if @xmath14 is a collection of subsets of @xmath2 $ ] , each of which are pairwise weakly separated from each other , and such that @xmath14 is not contained in any larger collection with this property , then @xmath16 . the above notion of weak separation
is related to the study of the plcker coordinates on the flag manifold , see @xcite .
similarly , in the context of plcker coordinates on the grassmannian , it is natural to study weak separation of @xmath17 element subsets of @xmath2 $ ] , for fixed @xmath18 and @xmath17 . observe that , when @xmath19 , the definition of weak separation becomes invariant under cyclic shifts of @xmath2 $ ] . indeed , in this case @xmath0 and @xmath1 are weakly separated if and only if after an appropriate cyclic shift @xmath20 .
scott @xcite proved that , for any collection @xmath21}{k}$ ] of pairwise weakly separated @xmath17 element subsets on @xmath2 $ ] , one has @xmath22 . here
@xmath23}{k}$ ] denotes the set of @xmath17 element subsets of @xmath2 $ ] .
moreover , scott made the following conjecture .
@xcite [ conj : scott ] if @xmath24}{k}$ ] is a collection of @xmath17 element subsets of @xmath2 $ ] , each of which are pairwise weakly separated from each other , and such that @xmath14 is not contained in any larger collection with this property , then @xmath25 .
we will present a stronger statement , which implies both these conjectures . in @xcite , a stratification of the totally nonnegative grassmannian @xmath26 , called the _ positroid stratification , _
was investigated .
the cells of this stratification can be labeled by many different kinds of combinatorial objects .
the three of these objects which we will use are 1 .
grassmann necklaces , which are certain sequences @xmath27 in @xmath23}{k}$ ] .
positroids , which are certain subsets of @xmath23}{k}$ ] .
3 . decorated permutations , which are permutations of @xmath2 $ ] with fixed points colored in two colors .
there is a length function @xmath28 on decorated permutations , related to the length function on the affine symmetric group .
one of our main results is the following .
[ thm : main_intro ] let @xmath29 , @xmath30 and @xmath31 be a grassmann necklace , positroid and decorated permutation corresponding to each other .
let @xmath32}{k}$ ] be a collection of pairwise weakly separated sets such that @xmath33 , and such that @xmath14 is not contained in any larger collection with this property
. then @xmath34
. we will call collections @xmath14 satisfying the conditions of this theorem _ maximal weakly separated collections inside @xmath30 . _ this result implies conjectures [ conj : lz ] and [ conj : scott ] , and also the @xmath35-chamber conjecture of leclerc and zelevinsky @xcite . from our perspective ,
scott s conjecture [ conj : scott ] is especially natural , because it refers to the largest cell in @xmath36
. the reader may find it useful to first consider the proofs of the main results in that special case .
we also prove the following result on mutation - connectedness , whose special cases were conjectured in @xcite and @xcite .
[ thm : mutations_intro ] fix a positroid @xmath30 .
any two maximal weakly separated collections inside @xmath30 can be obtained from each other by a sequence of mutations of the following form : @xmath37 , assuming that , for some cyclically ordered elements @xmath38 in @xmath2 \setminus s$ ] , @xmath14 contains @xmath39 , @xmath40 , @xmath41 , @xmath42 and @xmath43 .
( here @xmath39 is a shorthand for @xmath44 , etc . )
our main tool is the technology of plabic graphs , developed in @xcite .
plabic graphs are variants of wiring diagrams of decompositions in the symmetric group , which deserve to be better understood . as with wiring diagrams , there is a notion of a reduced plabic graph , and the faces of a reduced plabic graph are labeled with subsets of @xmath23}{k}$ ] . actually , we show that weakly separated collections and reduced plabic graphs are in a bijective correspondence .
[ th : intro_correspondence ] fix a positroid @xmath30 and the corresponding grassmann necklace @xmath29 . for a reduced plabic graph @xmath45 associated with @xmath30 , let @xmath46}{k}$ ] be the collection of labels of faces of @xmath45 .
then the map @xmath47 is a bijection between reduced plabic graphs for the positroid @xmath30 and maximal weakly separated collections @xmath14 such that @xmath33 .
we also establish the following consequence for the theory of cluster algebras , in section [ sec : clust ] .
let @xmath14 be a subset of @xmath23}{k}$ ] .
the following are equivalent : 1 .
the set of plcker coordinates @xmath48 is a cluster in the cluster algebra structure on the coordinate ring of @xmath49 .
@xmath14 is a maximal weakly separated collection .
@xmath14 is the collection of face labels of a reduced plabic graph for the uniform matroid @xmath23}{k}$ ] .
readers not interested in cluster algebras may skip this section .
theorems [ thm : main_intro ] and [ thm : mutations_intro ] follow from the correspondence in theorem [ th : intro_correspondence ] and the properties of plabic graphs from @xcite .
our main new tool is a construction we term plabic tilings " .
maximal plabic tilings are dual to plabic graphs , and non - maximal plabic tilings correspond to non - maximal weakly separated collections . as we were completing this work , we learned that danilov , karzanov and koshevoy succeeded in proving leclerc and zelevinsky s conjectures .
like our proof , their proof relies on certain planar diagrams .
their diagrams , which are called generalized tilings , are closely related to the plabic graphs which play a central role in this paper , see @xcite .
we see several differences between our work and theirs .
we study weakly separated collections of all positroids , which is more general than considering weakly separated collections of @xmath35-chamber sets , see section [ sec : connection_with_lz ] .
moreover , our constructions respect the dihedral symmetry inherent in the definitions of positroids and weak separation , which danilov , karzanov and koshevoy s constructions do not .
finally , our bicolored surfaces give a natural object to assign to a weakly separated collection which is not maximal . from our perspective , it is most natural to work with weak separation as defined by scott @xcite , which is motivated by geometric considerations in the case of grassmannians .
a similar notion was introduced by leclerc and zelevinsky , @xcite , motivated by considerations in the study of flag manifolds .
we will refer to leclerc and zelevinsky s definition as * _ lz weak separation_*. as we explain in section [ sec : connection_with_lz ] , lz weak separation is actually a special case of weak separation .
all of the statements in section [ sec : connection_with_lz ] ( although not all their proofs ) can be understood as soon as the reader has read the definitional material in sections [ sec : ws ] and [ sec : grneck ] .
the third author is engaged in an as yet unpublished project with andre henriques and dylan thurston , which has some overlap with this work .
in particular , proposition [ prop : embed ] was originally proved in that collaboration ; the third author is grateful to his collaborators for letting it appear here .
throughout the paper we use the following notation . let @xmath2 $ ] denote @xmath50 , and @xmath23}{k}$ ] denote the set of @xmath17 element subsets of @xmath2 $ ] .
we will generally consider @xmath2 $ ] as cyclically ordered .
we write @xmath51 for the cyclically shifted linear order on
@xmath2 $ ] : @xmath52 we will say that @xmath53 , @xmath54 , , @xmath55 in @xmath2 $ ] are _ cyclically ordered _ if @xmath56 for some @xmath57 $ ] .
we write @xmath58 for the open cyclic interval from @xmath59 to @xmath60 . in other words ,
the set of @xmath61 such that @xmath59 , @xmath61 , @xmath60 is cyclically ordered .
we write @xmath62 $ ] for the closed cyclic interval , @xmath63 = ( a , b ) \cup \ { a , b \}$ ] , and use similar notations for half open intervals .
if @xmath64 is a subset of @xmath2 $ ] and @xmath59 an element of @xmath2 $ ] , we may abbreviate @xmath65 and @xmath66 by @xmath67 and @xmath68 . in this paper
, we need to deal with three levels of objects : elements of @xmath2 $ ] , subsets of @xmath2 $ ] , and collections of subsets of @xmath2 $ ] . for clarity
, we will denote these by lower case letters , capital letters , and calligraphic letters , respectively .
the use of the notation @xmath4 does not imply @xmath69 .
in this section , we define weak separation for collections of @xmath17 element subsets and discuss the @xmath17 subset analogue of leclerc and zelevinsky s conjectures .
the relation of this approach to the original definitions and conjectures from @xcite will be discussed in section [ sec : connection_with_lz ] .
let us fix two nonnegative integers @xmath70 . for two @xmath17 element subsets @xmath0 and @xmath1 of @xmath2 $ ]
, we say that @xmath0 and @xmath1 are _ weakly separated _ if there do _ not _ exist @xmath59 , @xmath60 , @xmath71 , @xmath72 , cyclically ordered , with @xmath73 and @xmath74 .
geometrically , @xmath0 and @xmath1 are weakly separated if and only if there exists a chord separating the sets @xmath75 and @xmath76 drawn on a circle .
we write @xmath77 to indicate that @xmath0 and @xmath1 are weakly separated .
we call a subset @xmath14 of @xmath23}{k}$ ] a _ collection .
_ we define a _ weakly separated collection _ to be a collection @xmath21}{k}$ ] such that , for any @xmath0 and @xmath1 in @xmath14 , the sets @xmath0 and @xmath1 are weakly separated .
we define a _
maximal weakly separated collection _ to be a weakly separated collection which is not contained in any other weakly separated collection . following leclerc and zelevinsky @xcite , scott observed the following claim .
@xcite , cf .
@xcite [ mutation ] let @xmath78}{k-2}$ ] and let @xmath59 , @xmath60 , @xmath79 , @xmath80 be cyclically ordered elements of @xmath2 \setminus s$ ] .
suppose that a maximal weakly separated collection @xmath14 contains @xmath39 , @xmath81 , @xmath41 , @xmath42 and @xmath43 .
then @xmath82 is also a maximal weakly separated collection .
we define @xmath14 and @xmath83 to be _ mutations _ of each other if they are linked as in proposition [ mutation ] .
we will prove the following claim , conjectured by scott .
[ sconjecture ] every maximal weakly separated collection of @xmath23}{k}$ ] has cardinality @xmath84 .
any two maximal weakly separated collections are linked by a sequence of mutations .
while our results are purely combinatorial , they are motivated by constructions in algebraic geometry . specifically , @xcite introduced the positroid stratification of the grassmannian ( see also @xcite ) .
we will prove a version of theorem [ sconjecture ] for every cell in this stratification .
theorem [ sconjecture ] itself will correspond to the case of the largest cell .
there are several combinatorial objects which can be used to index the cells of this stratification .
we will use three of these _ grassmann necklaces , decorated permutations , _ and _ positroids_. see @xcite for more details .
* definition 16.1 ) a _ grassmann necklace _ is a sequence @xmath85 of @xmath17 element subsets of @xmath2 $ ] such that , for @xmath86 $ ] , the set @xmath87 contains @xmath88 .
( here the indices are taken modulo @xmath18 . )
if @xmath89 , then we should have @xmath90 .
in other words , @xmath87 is obtained from @xmath91 by deleting @xmath61 and adding another element , or @xmath90 .
note that , in the latter case @xmath92 , either @xmath61 belongs to all elements @xmath93 of the grassmann necklace , or @xmath61 does not belong to all elements @xmath93 of the necklace .
here is an example of a grassmann necklace : @xmath94 , @xmath95 , @xmath96 , @xmath97 , @xmath98 .
recall the linear order @xmath99 on @xmath2 $ ] .
we extend @xmath99 to @xmath17 element sets , as follows . for @xmath100 and @xmath101 with @xmath102 and @xmath103 , define the partial order @xmath104 in other words , @xmath105 is the cyclically shifted termwise partial order on @xmath23}{k}$ ] . given a grassmann necklace @xmath106 , define the _ positroid _ @xmath107 to be @xmath108}{k } \mid i_i \leq_i j \text { for all } i \in [ n ] \}.\ ] ] this turns out to be a special kind of matroid .
the term `` positroid '' is an abbreviation for `` totally positive matroid '' .
these are exactly matroids that can be represented by totally positive matrices , see @xcite .
fix a grassmann necklace @xmath106 , with corresponding positroid @xmath107 .
then @xmath14 is called a _ weakly separated collection inside _
@xmath107 if @xmath14 is a weakly separated collection and @xmath109 .
we call @xmath14 a _ maximal weakly separated collection inside _ @xmath107 if it is maximal among weakly separated collections inside @xmath107 .
we have not shown yet that @xmath110 , or that @xmath29 is weakly separated , so there is a risk that there are no weakly separated collections in @xmath107 .
we remedy this by the following lemma .
[ lem : iinm ] for any grassmann necklace @xmath29 , we have @xmath110 , and @xmath29 is weakly separated . for every @xmath61 and @xmath111 in @xmath2
$ ] , we must show that @xmath112 and @xmath113 . by the definition ,
@xmath114 is either obtained from @xmath115 by deleting @xmath116 and adding another element , or else @xmath117 . as we do the changes @xmath118 , we delete each @xmath119 $ ] at most once ( in the transformation @xmath120 ) .
this implies that we add each @xmath116 at most once .
let us show that @xmath121 .
suppose that this is not true and there exists @xmath122 .
( note that @xmath123 .
otherwise , @xmath116 belongs to all elements of the grassmann necklace , or @xmath116 does not belong to all elements of the necklace . )
consider the sequence of changes @xmath124 .
we should have @xmath125 , @xmath126 , @xmath127 , @xmath128 .
thus @xmath116 should be added twice , as we go from @xmath91 to @xmath115 and as we go from @xmath114 to @xmath93 .
we get a contradiction .
thus @xmath121 and , similarly , @xmath129 .
we conclude that @xmath112 and @xmath113 , as desired .
our main theorem will say , in part , that all maximal weakly separated collections inside @xmath107 have the same size . to describe this cardinality , we define decorated permutations .
* definition 13.3 ) a decorated permutation @xmath130 is a permutation @xmath131 together with a coloring function @xmath132 from the set of fixed points @xmath133 to @xmath134 .
there is a simple bijection between decorated permutations and grassmann necklaces . to go from a grassmann necklace @xmath29 to a decorated permutation @xmath135
, we set @xmath136 whenever @xmath137 for @xmath138 . if @xmath139 then @xmath140 is a fixed point of color @xmath141 .
finally , if @xmath142 then @xmath140 is a fixed point of color @xmath143 . to go from a decorated permutation @xmath135 to a grassmann necklace @xmath29 , we set @xmath144 \mid j < _ i \pi^{-1}(j ) \textrm { or } ( \pi(j)=j \textrm { and } col(j)=-1 ) \}.\ ] ] for example , the decorated permutation @xmath145 with @xmath146 and @xmath147 corresponds to the grassmann necklace @xmath148 with @xmath149 , @xmath150 , @xmath151 , @xmath152 , @xmath153 , @xmath154 , @xmath155 , @xmath156 .
for the fact that this is a bijection between grassmann necklaces and decorated permutations , see ( * ? ? ?
* theorem 17.1 ) .
* section 17 ) for @xmath61 , @xmath157 $ ] , we say that @xmath158 forms an _ alignment _ in @xmath31 if @xmath61 , @xmath159 , @xmath160 , @xmath111 are cyclically ordered ( and all distinct ) . the _ length _
@xmath161 is defined to be @xmath162 where @xmath163 is the number of alignments in @xmath31 .
we define @xmath164 to be @xmath161 where @xmath165 is the associated decorated permutation of @xmath29 .
we now state our result for an arbitrary grassmann cell .
[ thm : gsconjecture ] fix any grassmann necklace @xmath29 .
every maximal weakly separated collection inside @xmath107 has cardinality @xmath166 .
any two maximal weakly separated collections inside @xmath107 are linked by a sequence of mutations . as a particular case ,
let @xmath167 $ ] .
( the entries of @xmath91 are taken modulo @xmath18 . )
then @xmath168}{k}$ ] and the above theorem becomes theorem [ sconjecture ] . in section [ sec : connection_with_lz ] , we explain how this theorem also incorporates the conjectures of leclerc and zelevinsky .
let @xmath29 be a grassmann necklace , let @xmath169 be the corresponding decorated permutation , and let @xmath30 be the corresponding positroid @xmath107 .
the connected components of @xmath169 , @xmath29 , and @xmath30 are certain decorated permutations , grassmann necklaces , and positroids , whose ground sets may no longer be @xmath2 $ ] but rather subsets of @xmath2 $ ] , which inherit their circular order from @xmath2 $ ] . the arrows in the left hand side of figure [ fig : disconnected ] form a decorated permutation with @xmath170 .
the connected components are @xmath171 , @xmath172 and @xmath173 .
the @xmath174 element set drawn on the faces of this graph form a weakly separated collection for this @xmath31 .
the reader who is curious about the bipartite graph in figure [ fig : disconnected ] , and about the right hand side of the figure , will have her curiosity satisfied in sections [ sec : plabicgraphs ] and [ sec : tilings ] .
let @xmath2 = s_1 \sqcup s_2 \sqcup \cdots \sqcup s_r$ ] be a partition of @xmath2 $ ] into disjoint subsets .
we say that @xmath2 $ ] is * _ noncrossing _ * if , for any circularly ordered @xmath175 , we have @xmath176 and @xmath177 then @xmath178 .
see , for example , @xcite for background on noncrossing partitions .
the following is obvious from the definition : the common refinement of two noncrossing partitions is noncrossing .
therefore , the following definition makes sense .
let @xmath179 be a decorated permutation .
let @xmath2 = \bigsqcup s_i$ ] be the finest noncrossing partition of @xmath2 $ ] such that , if @xmath180 then @xmath181 .
let @xmath182 be the restriction of @xmath169 to the set @xmath183 , and let @xmath184 be the associated grassmann necklace on the ground set @xmath183 , for @xmath185 .
we call @xmath182 the _ connected components _ of @xmath169 , and @xmath184 the _ connected components _ of @xmath29 .
we say that @xmath169 and @xmath29 are connected if they have exactly one connected component .
[ lem : twointervals ] the decorated permutation @xmath179 is disconnected if and only if there are two circular intervals @xmath186 and @xmath187 such that @xmath31 takes @xmath186 and @xmath187 to themselves .
if such intervals exist , then the pair @xmath2 = [ i , j ) \sqcup [ j , i)$ ] is a noncrossing partition preserved by @xmath31 .
so there is a nontrivial noncrossing partition preserved by @xmath31 and @xmath31 is not connected .
conversely , any nontrivial noncrossing permutation can be coarsened to a pair of intervals of this form so , if @xmath31 is disconnected , then there is a pair of intervals of this form .
note that each fixed point of @xmath169 ( of either color ) forms a connected component .
a grassmann necklace @xmath188 is connected if and only if the sets @xmath189 are all distinct .
if @xmath179 is disconnected then let @xmath186 and @xmath187 be as in lemma [ lem : twointervals ] .
as we change from @xmath91 to @xmath87 to @xmath190 to to @xmath93 , each element of @xmath186 is removed once and is added back in once .
so @xmath191 .
conversely , suppose that @xmath191 .
as we change from @xmath91 to @xmath87 to @xmath190 and so forth , up to @xmath93 , each element of @xmath186 is removed once . in order to have @xmath191
, each element of @xmath186 must be added back in once .
so @xmath31 takes @xmath186 to itself . in this section
, we will explain how to reduce computations about positroids to the connected case .
so , for the rest of this section , suppose that @xmath191 for some @xmath192 . set @xmath193 and @xmath194 ; set @xmath195 and @xmath196 .
we will also write @xmath197 and @xmath198 .
[ directsum ] for every @xmath199 , we have @xmath200 and @xmath201 .
since @xmath91 is the @xmath105 minimal element of @xmath30 , we have @xmath202 for all @xmath199 . but
also , similarly , @xmath203 adding these inequalities together , we see that @xmath204 but , in fact , @xmath205 .
so we have equality at every step of the process .
in particular , @xmath200 and @xmath206 .
[ p : itsdirectsum ] the matroid @xmath30 is a direct sum of two matroids @xmath207 and @xmath208 , supported on the ground sets @xmath186 and @xmath187 , having ranks @xmath209 and @xmath210 . in other words , there are matroids @xmath207 and @xmath208 such that @xmath1 is in @xmath30 if and only if @xmath211 is in @xmath207 and @xmath212 is in @xmath208 .
for every @xmath199 , write @xmath213 and @xmath214 .
this is an immediate corollary of proposition [ directsum ] ; see ( * ? ? ?
* theorem 7.6.4 ) for the fact that this rank condition implies that @xmath30 is the direct sum of @xmath215 and @xmath216 .
see ( * ? ? ?
* proposition 7.6.1 ) for the interpretation of this in terms of bases of @xmath30 . for @xmath217 $ ] , the set @xmath218 is of the form @xmath219 for some @xmath220 . for @xmath221 $ ] ,
the set @xmath218 is of the form @xmath222 for some @xmath223 .
consider the case that @xmath217 $ ] , the other case is similar .
recall that @xmath218 is the @xmath224 minimal element of @xmath30 . since @xmath225 , we know that @xmath226 , where @xmath227 is the @xmath224 minimal element of @xmath207 .
but , on @xmath187 , the orders @xmath105 and @xmath224 coincide , so @xmath228 is the @xmath105 minimal element of @xmath208 , namely @xmath229 .
view @xmath186 as circularly ordered .
let @xmath230 denote the circularly ordered sequence @xmath231 .
@xmath230 is a grassmann necklace on the ground set @xmath186 , and @xmath207 is the associated positroid . for @xmath232 , we know that @xmath233 . since @xmath234 , and @xmath235 , we have @xmath236 .
this is the definition of a grassmann necklace .
( note that we have used the condition @xmath191 to cover the boundary case @xmath237 . )
now , we show that @xmath30 is the associated positroid .
consider any @xmath238 . if @xmath239 then @xmath240 , so @xmath241 for every @xmath242 $ ] .
this immediately implies that @xmath243 for every @xmath232 , so @xmath1 is in the positroid @xmath244 .
conversely , suppose that @xmath1 is in the positroid @xmath244 .
we wish to show that @xmath219 is in @xmath30 . reversing
the argument of the previous paragraph shows that @xmath241 for all @xmath232 . for @xmath245 , we know that @xmath246 and , since @xmath247 and @xmath248 coincide on @xmath186 , we know that @xmath249 .
so @xmath241 for @xmath17 in @xmath187 as well .
so @xmath199 .
it is easy to check the following : with the above definitions , @xmath250 .
we now study weakly separated collections in @xmath30 .
[ splitlemma ] let @xmath251 .
if @xmath1 is weakly separated from @xmath252 , then either @xmath253 or @xmath254 .
suppose , to the contrary , that @xmath255 and @xmath256 .
since @xmath257 , there are @xmath59 and @xmath258 , with @xmath259 , such that @xmath260 and @xmath261 . similarly , there are @xmath79 and @xmath262 , with @xmath263 , such that @xmath264 and @xmath265
. then @xmath59 and @xmath79 are in @xmath266 , while @xmath60 and @xmath80 are in @xmath267 .
so @xmath252 and @xmath1 are not weakly separated .
[ splittingsummary ] if @xmath14 is a weakly separated collection in @xmath30 , then there are weakly separated collections @xmath268 and @xmath269 in @xmath207 and @xmath208 such that @xmath270 conversely , if @xmath268 and @xmath269 are weakly separated collections in @xmath207 and @xmath208 , then the above formula defines a weakly separated collection in @xmath30 .
the collection @xmath14 is maximal if and only if @xmath268 and @xmath269 are .
first , suppose that @xmath14 is a weakly separated collection in @xmath30 . since @xmath271 , we have @xmath272 . by lemma [ splitlemma ] ,
every @xmath273 is either of the form @xmath274 , or @xmath275 .
let @xmath276 be the collection of all sets @xmath227 for which @xmath277 is in @xmath14 .
the condition that @xmath14 is weakly separated implies that @xmath276 is ; the condition that @xmath33 implies that @xmath278 .
so @xmath276 is a weakly separated collection in @xmath279 and it is clear that @xmath14 is built from @xmath268 and @xmath269 in the indicated manner .
conversely , it is easy to check that , if @xmath268 and @xmath269 are weakly separated collections in @xmath207 and @xmath208 , then the above formula gives a weakly separated collection in @xmath30 . finally , if @xmath280 with @xmath83 a weakly separated collection in @xmath30 , then either @xmath281 or @xmath282 .
so , if @xmath14 is not maximal , either @xmath268 or @xmath269 is not .
the converse is similar . in summary
, the grassmann necklace @xmath29 can be described in terms of smaller necklaces @xmath230 and @xmath283 ; the positroid @xmath30 can be described in terms of smaller positroids @xmath207 and @xmath208 ; weakly separated collections @xmath14 in @xmath30 can be described in terms of weakly separated collections in @xmath207 and @xmath208 . the connected components @xmath284 are the connected components of @xmath30 in the sense of matroid theory .
we have partially proved this in this section . in order to fully establish this result
we would also need to show that , if @xmath179 is connected , then the matroid @xmath30 is connected .
we do nt need this fact , but it is true .
plabic graphs were introduced in @xcite .
we will use these graphs to represent weakly separated collections and the properties of these graphs will be a major key in our proof of the conjecture . for more details on plabic graphs ,
see @xcite . a * _
planar bicolored graph _ * , or simply a * _ plabic graph _ * is a planar undirected graph @xmath45 drawn inside a disk .
the vertices on the boundary are called boundary vertices , and are labeled in clockwise order by @xmath2 $ ] .
all vertices in the graph are colored either white or black .
let @xmath45 be a plabic graph in the disc @xmath285 .
we will draw @xmath18 directed paths , called * _ strands _ * , within the disc @xmath285 , each starting from and ending at a boundary vertex of @xmath45 .
the construction in this definition is depicted in figure [ fig : pla_label ] ; the numeric labels in that figure will be explained below .
the strands are drawn as follows : for each edge of @xmath45 , draw two strand segments .
if the ends of the segment are the same color , then the two strands should be parallel to the edge without crossing , and should run in opposite directions . if the two ends are different colors , then the two strands should cross , with one running towards each endpoint .
as we will discuss in the below remark , for most purposes , we can reduce to the case that @xmath45 is bipartite , so the latter case is the important one . around each vertex ,
connect up the ends of the strands so that they turn right at each black vertex and left at each white vertex .
we will have @xmath18 strands leading from @xmath286 to itself , plus possibly some loops in the interior of @xmath45 .
[ rem : contract ] suppose that @xmath287 and @xmath288 are two vertices of @xmath45 , joined by an edge , which have the same color .
let @xmath289 be the graph formed by contracting the edge @xmath290 to a single vertex @xmath35 , and coloring @xmath35 the same color as @xmath287 and @xmath288 are colored .
then the strands of @xmath45 and @xmath289 have the same connectivity .
see figure [ contraction ] for a depiction of how contracting an edge leaves strand topology unchanged . for this reason
, we can usually reduce any question of interest to the case where @xmath45 is bipartite .
we do not restrict in this paper to bipartite graphs for two reasons : ( 1 ) we want to be compatible with the definitions in @xcite , which does not make this restriction and ( 2 ) the description of the square move , ( m1 ) below , would be significantly more complicated .
a plabic graph is called * _ reduced _ * ( * ? ? ?
* section 13 ) if the following holds : 1 .
the strands can not be closed loops in the interior of the graph .
no strand passes through itself .
the only exception is that we allow simple loops that start and end at a boundary vertex @xmath61 .
[ noloop ] 3 . for any two strands @xmath291 and @xmath292 , if @xmath291 and @xmath292 have two common vertices @xmath12 and @xmath13 , then one strand , say @xmath291 , is directed from @xmath12 to @xmath13 , and the other strand @xmath292 is directed from @xmath13 to @xmath12 .
( that is the crossings of @xmath291 and @xmath292 occur in opposite orders in the two strands . )
[ oppositeorder ] strands , and the reducedness condition , occurs in the physics - inspired literature on quivers . in this literature
, strands are called zig - zag paths .
graphs which we call `` reduced '' are said to obey condition @xmath293 " in ( * ? ? ?
* section 8) and are called `` marginally geometrically consistent '' in ( * ? ? ?
* section 3.4 ) . the strand which _ ends _ at the boundary vertex @xmath61 is called strand @xmath61 .
* section 13 ) for a reduced plabic graph @xmath45 , let @xmath294 be the permutation such that the strand that starts at the boundary vertex @xmath61 ends at the boundary vertex @xmath295 .
a fixed point @xmath296 corresponds to simple loop at the boundary vertex @xmath61 .
we color a fixed point @xmath61 of @xmath297 as follows : @xmath141 if the corresponding loop is counter - clockwise ; and @xmath298 if the loop is clockwise . in this way
, we assign the * _ decorated strand permutation _ * @xmath299 to each reduced plabic graph @xmath45 .
we will label the faces of a reduced plabic graph with subsets of @xmath2 $ ] ; this construction was first published in @xcite . by condition [ noloop ]
each strand divides the disk into two parts . for each face
@xmath300 we label that @xmath300 with the set of those @xmath86 $ ] such that @xmath300 lies to the left of strand @xmath61 .
see figure [ fig : pla_label ] for an example .
so given a plabic graph @xmath45 , we define @xmath301 as the set of labels that occur on each face of that graph .
when we pass from one face @xmath300 of @xmath45 to a neighboring one @xmath302 , we cross two strands . for one of these strands
, @xmath300 lies on its left and @xmath302 on the right ; for the other @xmath300 lie on the right and @xmath302 on the left .
so every face is labeled by the same number of strands as every other .
we define this number to be the * _ rank _ * of the graph ; it will eventually play the role of @xmath17 .
the following claim establishes a correspondence between maximal weakly separated collections in a positroid and reduced plabic graphs .
it describes maximal weakly separated collections as labeled sets @xmath301 of reduced plabic graphs .
[ thm : wsispg ] for a decorated permutation @xmath169 and the corresponding grassmann necklace @xmath303 , a collection @xmath14 is a maximal weakly separated collection inside the positroid @xmath107 if and only if it has the form @xmath304 for a reduced plabic graph with strand permutation @xmath169 .
in particular , a maximal weakly separated collection @xmath14 in @xmath23}{k}$ ] has the form @xmath305 for a reduced plabic graph @xmath45 with strand permutation @xmath306 $ ] .
we define the boundary face at @xmath61 to be the face touching the part of the disk between boundary vertices @xmath307 and @xmath61 in clockwise order .
let @xmath91 be the label of that face . at boundary point @xmath61 ,
the strand @xmath61 comes in and the strand @xmath159 leaves .
so @xmath87 is obtained from @xmath91 by deleting @xmath61 and adding in @xmath159 ; we deduce that @xmath308 is the grassmann necklace @xmath309 .
we now describe how to see mutations in the context of plabic graphs .
we have following 3 moves on plabic graphs .
( m1 ) pick a square with vertices alternating in colors , such that all vertices have degree @xmath310 .
we can switch the colors of all the vertices .
see figure [ fig : square_move ] .
[ cc][cc]@xmath311 ( m2 ) for two adjoint vertices of the same color , we can contract them into one vertex .
see figure [ fig : edge_contraction ] .
[ cc][cc]@xmath311 ( m3 ) we can insert or remove a vertex inside any edge .
see figure [ fig : vertex_removal ] .
[ cc][cc]@xmath311 the moves do not change the associated decorated permutation of the plabic graph , and do not change whether or not the graph is reduced .
the power of these moves is reflected in the next theorem : ( * ? ? ?
* theorem 13.4 ) [ thm : p ] let @xmath45 and @xmath289 be two reduced plabic graphs with the same number of boundary vertices .
then the following claims are equivalent : * @xmath45 can be obtained from @xmath289 by moves ( m1)-(m3 ) . *
these two graphs have the same decorated strand permutation @xmath312 . moves ( m2 ) and ( m3 ) do not change @xmath301 , while move ( m1 ) changes @xmath301 by a mutation .
so , once we prove theorem [ thm : wsispg ] , we will establish that all maximal weakly separated collections within @xmath313 are connected to each other by mutations .
notice also that the moves ( m1)-(m3 ) do not change the number of faces of the plabic graph .
so all reduced plabic graphs for a given decorated permutation have the same number of faces .
this number was calculated in @xcite : [ thm : p2 ] let @xmath45 be a reduced plabic graph with decorated permutation @xmath165 .
then @xmath45 has @xmath314 faces . by (
* proposition 17.10 ) , @xmath315 is the dimension of @xmath316 , a manifold whose definition we do not need to know . by
( * ? ? ? * theorem 12.7 ) , @xmath316 is isomorphic to @xmath317 .
thus , proving theorem [ thm : wsispg ] will establish all parts of [ thm : gsconjecture ] , thus proving the conjectures of scott and of leclerc and zelevinsky .
this section is not cited in the rest of the paper . from the beginnings of the theory of cluster algebras ,
it has been expected that the coordinate ring of the grassmannian , in its plcker embedding , would be a cluster algebra , and that the plcker coordinates would be cluster variables .
this was verified by scott in @xcite .
moreover , scott showed that , given any reduced alternating strand diagram , the face labels of that diagram form a cluster in the cluster structure for @xmath49 . with the tools of this paper , we can establish the converse statement . [ cluster ]
let @xmath14 be a subset of @xmath23}{k}$ ] .
the following are equivalent : 1 .
the set of plcker coordinates @xmath48 is a cluster in the cluster algebra structure on the coordinate ring of @xmath49 .
@xmath14 is a maximal weakly separated collection .
@xmath14 is the collection of face labels of a reduced plabic graph for the uniform matroid @xmath23}{k}$ ] . in this section ,
we use the language of cluster algebras freely .
the implication @xmath318 is the main result of this paper ; the implication @xmath319 is ( * ? ? ?
* theorem 2 ) .
we now show @xmath320 .
let @xmath0 and @xmath273 ; we will show that @xmath77 .
let @xmath321 be the quantization of the coordinate ring of @xmath49 introduced in @xcite ; since @xmath322 and @xmath323 are in a common cluster , they quasi - commute in this ring . by @xcite
, the ring @xmath321 is isomorphic to the ring of quantum minors " of a @xmath324 matrix ( see @xcite or @xcite ) , so the quantum minors @xmath325 and @xmath326 quasi - commute . as computed in @xcite , if @xmath325 and @xmath326 quasi - commute , then @xmath0 and @xmath1 are weakly separated .
so we know that @xmath14 is a weakly separated collection .
every cluster has cardinality @xmath327 , so @xmath14 is a maximal weakly separated collection by the bound of @xcite .
we expect that an analogous statement holds for all positroids .
we have not proved it here for two reasons .
the first is that , although positroid varieties are widely expected to have a cluster structure , this has not yet been verified in print .
the second is that , since the coordinate ring of a positroid variety is a quotient of the coordinate ring of @xmath49 , it is possible that two minors which do not quasi - commute on @xmath49 do quasi - commute when restricted to this smaller subvariety . deriving the analogues of @xcite and @xcite for positroid varieties strikes us as an excellent project
our goal in this section is to show that , for any reduced plabic graph @xmath45 with associated decorated permutation @xmath169 , the collection @xmath301 is a weakly separated collection for @xmath328 .
this is the easy part of theorem [ thm : wsispg ] .
we will show that this collection is maximal in section [ sec : tilings ] , and that any maximal weakly separated collection is of the form @xmath301 in section [ sec : final ] .
[ prop : plagws ] let @xmath45 be a reduced plabic graph . then @xmath301 is a weakly separated collection .
assume @xmath301 is not weakly separated .
pick @xmath0 and @xmath329 such that @xmath330 .
let @xmath59 , @xmath60 , @xmath79 and @xmath80 be cyclically ordered elements such that @xmath331 and @xmath332 .
we first consider the case where strands @xmath59 and @xmath79 do nt cross ( see figure [ dontcross ] ) . in this case , region @xmath0 is to the left of strands @xmath59 and @xmath79 , and region @xmath1 is to their right , so strands @xmath59 and @xmath79 must be parallel , not antiparallel . in other words , the endpoints @xmath333 are circularly ordered . if @xmath334 is in @xmath335 , then strands @xmath59 and @xmath60 can not cross and @xmath0 is on the left side of @xmath60 , contrary to our desires .
if @xmath334 is in @xmath336 then strand @xmath60 can not cross strand @xmath79 and we deduce that @xmath1 is to the right of strand @xmath60 , again contrary to our desires .
but the intervals @xmath335 and @xmath337 cover all of @xmath2 $ ] , so this excludes all possible positions for @xmath334 and we have a contradiction .
this concludes the proof in the case that strands @xmath59 and @xmath79 do not cross .
[ cc][cc]@xmath0 [ cc][cc]@xmath1 [ cc][cc]@xmath59 [ cc][cc]@xmath79 now , suppose that strands @xmath59 and @xmath79 cross ( see figure [ lemma71 ] ) . let @xmath338 be the region which is to the left of @xmath59 and right of @xmath79 ; let @xmath339 be to the right of @xmath59 and left of @xmath79 , and let @xmath340 , @xmath341 ,
, @xmath342 be the regions which are both to the left of @xmath59 and @xmath79 or both to the right .
our numbering is such that strand @xmath59 first passes by @xmath340 and precedes in increasing order , while @xmath79 starts at @xmath342 and goes in decreasing order .
let face @xmath0 be in @xmath284 and @xmath1 be in @xmath183 ; note that @xmath192 .
we will discuss the case that @xmath343 ; the case that @xmath344 is equivalent by relabeling @xmath175 as @xmath345 .
[ cc][cc]@xmath0 [ cc][cc]@xmath1 [ cc][cc]@xmath338 [ cc][cc]@xmath340 [ cc][cc]@xmath341 [ cc][cc]@xmath346 [ cc][cc]@xmath347 [ cc][cc]@xmath339 [ cc][cc]@xmath59 [ cc][cc]@xmath60 [ cc][cc]@xmath79 [ cc][cc]@xmath80 [ cc][cc ] consider strand @xmath80 , as it passes through the various regions @xmath338 , @xmath340 , , @xmath342 and @xmath339 .
( in figure [ lemma71 ] , a portion of strand @xmath80 is shown as a dotted line . ) we claim that it is impossible that strand @xmath80 both passes from @xmath338 to @xmath348 to @xmath339 and from @xmath339 to @xmath349 to @xmath338 .
the proof is simple : if it did , either its crossings with strand @xmath59 or its crossings with @xmath79 would violate condition [ oppositeorder ] in the definition of reducedness .
on the other hand , strand @xmath80 must enter both @xmath338 and @xmath339 , as otherwise regions @xmath0 and @xmath1 would lie on the same side of strand @xmath80 .
let @xmath350 be the index such that strand @xmath80 travels from @xmath338 to @xmath351 to @xmath339 , or else from @xmath339 to @xmath351 to @xmath338 .
we must have @xmath352 , since @xmath1 and @xmath0 are opposite sides of @xmath80 , and @xmath80 must travel from @xmath338 to @xmath351 to @xmath339 , as @xmath0 is on the right of @xmath80 and @xmath1 on the left .
but the boundary point @xmath80 , where strand @xmath80 terminates , is in the cyclic interval @xmath353 .
so @xmath80 can not end in @xmath339 .
it also can not end in @xmath338 , as it goes from @xmath338 to @xmath351 to @xmath339 .
so it must end in @xmath340 or @xmath342 .
we discuss the former case , since the latter case is similar .
if @xmath80 ends in @xmath340 , it must come from @xmath339 , passing through @xmath59 .
but then the intersections of @xmath79 and @xmath80 violate condition [ oppositeorder ] .
we now use proposition [ prop : plagws ] , together with theorem [ thm : p ] , to demonstrate some facts about the face labels of plabic graphs .
now we want to show that @xmath354 is actually a weakly separated collection of @xmath328 .
we need the following proposition .
[ prop : reduction ] let @xmath45 be a reduced plabic graph with decorated permutation @xmath169 such that @xmath355 .
there is some pair of strands @xmath356 in @xmath165 which only crosses once , at an edge @xmath357 of @xmath45 , and such that , if we delete the edge @xmath357 , thus uncrossing the strands , the resulting graph @xmath289 is still reduced . according to ( * ? ? ? * lemma 18.9 ) , a _
removable _ edge has these properties .
( see @xcite for definitions . )
according to ( * ? ? ?
* corollary 18.10 ) , the removable edges of @xmath45 are in bijection with the decorated permutations covered by @xmath179 . since @xmath358 , it covers at least one decorated permutation .
[ prop : partconverse ] if @xmath45 is a reduced plabic graph with decorated permutation @xmath169 , then the following properties hold : 1 .
the boundary cells of @xmath45 are labeled by @xmath359 .
2 . every face of @xmath45 receives a separate label in @xmath354 .
@xmath354 is contained in @xmath328 .
we are going to use induction on @xmath360 .
when @xmath361 , any reduced plabic graph has only one face due to theorem [ thm : p2 ] .
this is possible only when @xmath31 is the identity .
then there is only one face , and it has the label @xmath362 .
this covers the base case for our induction argument .
let @xmath61 , @xmath111 , @xmath357 and @xmath289 be as in proposition [ prop : reduction ] .
by induction hypothesis , all the properties hold for @xmath289 .
we will use @xmath363 to denote the decorated permutation of @xmath289 and use @xmath364 for the grassmann necklace of @xmath289 .
we start with the first property .
swapping @xmath61 and @xmath111 going from @xmath365 to @xmath169 corresponds to changing @xmath366 to @xmath29 in a way that , if some @xmath367 contains only one of @xmath61 and @xmath111 , than it gets swapped with the other .
when going from @xmath289 to @xmath45 , the label of a boundary face gets changed only when it contains exactly one of @xmath61 and @xmath111 , and the change is by swapping one with the other .
so the boundary cells of @xmath45 are labeled by @xmath359 .
now we check the second property . since @xmath61 and @xmath111 only cross once , if two faces are assigned the same label in @xmath45 , that means they have to be in the same region with respect to strands @xmath61 and @xmath111 . but
this also means they are assigned the same label in @xmath289 , a contradiction .
we prove the third property directly .
assume for the sake of contradiction that we have some @xmath368 such that @xmath369 .
for @xmath370 and @xmath371 , let @xmath372 denote the first position such that @xmath373 . from this , we get @xmath374 , and hence @xmath375 . and for any @xmath376 $ ] , we have @xmath377 , and it follows from comparing the strands @xmath378 and @xmath379 that @xmath1 should contain @xmath379 . but
this implies that @xmath380 , a contradiction .
combining the first and third property of proposition [ prop : partconverse ] with proposition [ prop : plagws ] , we achieve the following claim : [ cor : partconverse ] if @xmath45 is a reduced plabic graph with boundary @xmath29 , then @xmath301 is a weakly spearated collection in @xmath107 .
we have @xmath381 by proposition [ prop : partconverse ] ; that @xmath29 is weakly separated by proposition [ prop : plagws ] ; and that @xmath382 by proposition [ prop : partconverse ] .
we have not yet shown that @xmath301 is maximal ; that will be theorem [ thrm : ismaximal ] .
given a maximal weakly separated collection @xmath14 of @xmath23}{k}$ ] , we need to construct a plabic graph @xmath45 such that @xmath383 . to do so , we will define a * _ plabic tiling_*. for a weakly separated collection @xmath14 , we will construct a @xmath384-dimensional cw - complex embedded in @xmath385 , and denote this complex by @xmath386 .
maximal plabic tilings will turn out to be dual to reduced bipartite plabic graphs .
we will associate a plabic tiling to any weakly separated collection , maximal or not .
see example [ incompleteexample ] for a discussion of @xmath386 for a non - maximal @xmath14 .
figure [ tilingexample ] shows @xmath386 for @xmath14 the collection of face labels in figure [ fig : pla_label ] .
the points @xmath387 , which we will introduce soon , are shown in the lower right of figure [ tilingexample ] .
they are shown in the same scale as the rest of the figure .
let us fix @xmath14 , a weakly separated collection in @xmath23}{k}$ ] .
for @xmath0 and @xmath388}{k}$ ] , say that @xmath0 * _ neighbors _ * @xmath1 if @xmath389 let @xmath390 be any @xmath391 element subset of @xmath2 $ ] .
we define the * _ white clique _ * @xmath392 to be the set of @xmath393 such that @xmath394 .
similarly , for @xmath395 a @xmath396 element subset of @xmath2 $ ] , we define the * _ black clique _ * @xmath397 for the set of @xmath393 which are contained in @xmath395 .
we call a clique * _ nontrivial _ * if it has at least three elements . observe that , if @xmath398 is a nontrivial clique , then it can not be both black and white .
let @xmath45 be a reduced plabic graph of rank @xmath17 .
the black ( respectively white ) cliques of @xmath301 correspond to the black ( white ) vertices of @xmath45 .
more precisely , for each vertex @xmath288 , the labels of the faces bordering @xmath288 form a clique , and all cliques are of this form .
observe that a white clique @xmath392 is of the form @xmath399 for some @xmath400 , @xmath401 ,
, @xmath402 , which we take to be cyclically ordered .
similarly , @xmath397 is of the form @xmath403 , with the @xmath404 s cyclically ordered .
if @xmath392 is nontrivial , we define the boundary of @xmath392 to be the cyclic graph @xmath405 similarly , the boundary of a nontrivial @xmath397 is @xmath406 if @xmath407 is a two element clique , then we define its boundary to be the graph with a single edge @xmath408 ; we define an one element clique to have empty boundary .
[ lem : nbhr ] let @xmath0 neighbor @xmath1 ; set @xmath409 and @xmath410 . if @xmath392 and @xmath397 are nontrivial , then there is an edge between @xmath0 and @xmath1 in the boundaries of @xmath392 and @xmath397 .
let @xmath61 be the lone element of @xmath4 and @xmath111 the lone element of @xmath8 .
let @xmath411 and @xmath412 .
so @xmath61 and @xmath111 occur among the @xmath59 s , and among the @xmath60 s .
consider the four sets : @xmath413 our goal is to show that either @xmath414 or @xmath415 .
suppose ( for the sake of contradiction ) that @xmath340 and @xmath347 are both nonempty , with @xmath416 and @xmath417 .
set @xmath418 and @xmath419 .
then @xmath420 , @xmath421 , @xmath422 and @xmath423 , so @xmath424 and @xmath425 are not weakly separated .
we have a contradiction and we deduce that at least one of @xmath340 and @xmath347 is empty . similarly , at least one of @xmath341 and @xmath346 is empty . on the other hand ,
@xmath392 is nontrivial , so at least one of @xmath340 and @xmath341 is nonempty .
similarly , at least one of @xmath346 and @xmath347 is nonempty .
since at least one of @xmath426 is empty , and at least one of @xmath427 is , we deduce that at least two of @xmath428 are empty .
similarly , at most two of them are empty , so precisely two of the @xmath284 s are empty .
checking the @xmath429 possibilities , the ones consistent with the above restrictions are that either @xmath340 and @xmath346 are empty , and @xmath341 and @xmath347 are not , or vice versa .
we now define a two dimensional cw - complex @xmath386 .
the vertices of @xmath386 will be the elements of @xmath14 .
there will be an edge @xmath430 if 1 .
@xmath431 is nontrivial and @xmath430 appears in the boundary of @xmath431 or 2 .
@xmath432 is nontrivial and @xmath430 appears in the boundary of @xmath432 or 3 .
@xmath433 .
there will be a ( two - dimensional ) face of @xmath386 for each nontrivial clique @xmath398 of @xmath14 .
the boundary of this face will be the boundary of @xmath398 . by lemma
[ lem : nbhr ] , all edges in the boundary of @xmath398 are in @xmath386 , so this makes sense
. we will refer to each face of @xmath386 as * _ black _ * or * _ white _ * , according to the color of the corresponding clique .
we call a cw - complex of the form @xmath386 a * _ plabic tiling_*. so far , @xmath386 is an abstract cw - complex .
our next goal is to embed it in a plane .
fix @xmath18 points @xmath434 , @xmath435 ,
, @xmath436 in @xmath385 , at the vertices of a convex @xmath18-gon in clockwise order .
define a linear map @xmath437 by @xmath438 . for @xmath439}{t}$ ] , set @xmath440 .
we will need a notion of weak separation for vectors in @xmath441}$ ] .
let @xmath357 and @xmath442 be two vectors in @xmath441}$ ] , with @xmath443 } e_{a } = \sum_{a \in [ n ] } f_{a}$ ] .
we define @xmath357 and @xmath442 to be * _ weakly separated _ * if there do _ not _ exist @xmath59 , @xmath60 , @xmath71 and @xmath72 , cyclically ordered , with @xmath444 , @xmath445 , @xmath446 and @xmath447 .
so , for @xmath0 and @xmath388}{k}$ ] , we have @xmath77 if and only if @xmath448 .
[ lem : vecinj ] let @xmath357 and @xmath442 be two different points in @xmath449 , with @xmath443 } e_{a } = \sum_{a \in [ n ] } f_{a}$ ] and @xmath450
. then @xmath451 .
since @xmath357 and @xmath442 are weakly separated , there are some @xmath28 and @xmath116 in @xmath2 $ ] such that @xmath452 is nonnegative for @xmath59 in @xmath453 and is nonpositive for @xmath59 in @xmath454 .
we have @xmath455 since @xmath443 } e_{a } = \sum_{a \in [ n ] } f_{a}$ ] , the right hand side of the above equation is a positive linear combination of vectors of the form @xmath456 , with @xmath457 and @xmath458 .
since the @xmath459 s are the vertex of a convex @xmath18-gon , there is a line @xmath460 in @xmath385 separating @xmath461 from @xmath462 .
so every vector @xmath456 as above crosses from the former side of @xmath460 to the latter .
a positive linear combination of such vectors must cross the line @xmath460 , and can therefore not be zero .
so the right hand side of the above equation is nonzero , and we deduce that @xmath451 .
we extend the map @xmath31 to a map from @xmath386 to @xmath385 as follows : each vertex @xmath0 of @xmath386 is sent to @xmath463 and each face of @xmath386 is sent to the convex hull of the images of its vertices .
we encourage the reader to consult figure [ tilingexample ] and see that the vector @xmath464 is a translation of @xmath465 .
[ prop : embed ] for any weakly separated collection @xmath14 , the map @xmath31 embeds the cw - complex @xmath386 into @xmath385 .
suppose , to the contrary , that there are two faces of @xmath386 whose interiors have overlapping image .
we will deal with the case that both of these faces are two dimensional , with one white and the other black .
the other possibilities are similar , and easier .
let the vertices of the two faces be @xmath399 and @xmath466 .
write @xmath467 and @xmath468 . before we analyze the geometry of @xmath31
, we will need to do some combinatorics .
note that , if @xmath469 , then the faces are convex polygons with a common edge , lying on opposite sides of that edge , and thus have disjoint interiors .
so we may assume that @xmath470 is nonempty .
also , @xmath471 , so @xmath472 is nonempty .
let @xmath473 and @xmath474 .
suppose that @xmath475 and @xmath476 are in @xmath477 .
then weak separation of @xmath478 and @xmath479 implies that @xmath480 .
more generally , if @xmath481 and @xmath482 are both in @xmath477 , we still have @xmath483 .
similarly , if @xmath481 and @xmath482 are both in @xmath484 , then @xmath485 .
so , there is some @xmath486 $ ] and @xmath487 $ ] such that @xmath488 and @xmath489 .
now , let @xmath490 be the point which is in the interior of both faces , say @xmath491 for some positive scalars @xmath492 and @xmath493 with @xmath494
. define the vectors @xmath287 and @xmath288 by @xmath495 and @xmath496 , so @xmath497 .
all the positive entries of @xmath498 are contained in @xmath499 , and all the negative entries in @xmath500 .
so @xmath287 and @xmath288 are weakly separated and , by lemma [ lem : vecinj ] , @xmath501 , a contradiction .
looking through the proof , we have proved the more technical result : [ writethis ] let @xmath424 and @xmath425 be different faces of @xmath386 .
let @xmath502 and @xmath503 be the sets of vertices of the faces @xmath424 and @xmath425 .
let @xmath287 be a vector of the form @xmath504 , where @xmath505 and @xmath506 .
similarly , let @xmath288 be a vector of the form @xmath507 , with @xmath508 and @xmath509 .
then @xmath287 and @xmath288 are weakly separated .
[ rem : circ ] let @xmath510 , @xmath511 ,
, @xmath512 be the vertices of a white face of @xmath14 , with the @xmath475 in cyclic order .
then the vertices @xmath513 appear in clockwise order in the planar embedding . if the vertices of a black face are @xmath514 , @xmath515 ,
, @xmath516 , with the @xmath404 again in cyclic order , then the vertices @xmath517 again appear in clockwise order .
this is because negation is an orientation preserving operation on @xmath385 ; see figure [ tilingexample ] .
now , we will study the construction @xmath386 when @xmath14 is a weakly separated collection for a particular positroid @xmath30 .
propositions [ glueatpoint ] through [ lem : intervals ] are obvious in the case that @xmath30 is the largest positroid , @xmath23}{k}$ ] .
let @xmath30 be a positroid and @xmath29 the corresponding grassmann necklace .
we first give a lemma to reduce to the case when @xmath29 is connected : suppose that @xmath29 is not connected , with @xmath191 .
let @xmath14 be a weakly separated collection for @xmath279 ; and use the notations @xmath276 , @xmath518 and so forth from section [ connected ] .
it is easy to check that : [ glueatpoint ] the complex @xmath386 is formed by gluing @xmath519 and @xmath520 together at the point @xmath252 .
see figure [ fig : disconnected ] for an example of a disconnected plabic graph and the corresponding plabic tiling .
we now , therefore , restrict to the case that @xmath29 is a connected grassmann necklace .
we define @xmath521 to be the closed polygonal curve whose vertices are , in order , @xmath522 , @xmath523 , , @xmath524 , @xmath522 .
the curve @xmath521 is a simple closed curve . when @xmath30 is @xmath23}{k}$ ] , the interior of @xmath521 is convex , but this is not true in general .
suppose for the sake of contradiction that the curve @xmath521 crosses through itself .
first , by assumption , @xmath525 so , by lemma [ lem : vecinj ] , the vertices @xmath526 and @xmath527 are disjoint .
we now rule out the case that line segments @xmath528 and @xmath529 cross in their interior .
the case that @xmath528 passes through the vertex @xmath527 is similar and easier .
the collection @xmath29 is weakly separated by lemma [ lem : iinm ] . by proposition
[ prop : embed ] , the map @xmath31 embeds @xmath530 into @xmath385 .
the only way that the line segments @xmath528 and @xmath529 could cross is if @xmath91 , @xmath93 , @xmath87 and @xmath93 are vertices of some two - dimensional face of @xmath530 , arranged in that circular order .
( or the reverse circular order , but then we could switch the indices @xmath61 and @xmath111 . )
we consider the case that this face is of the form @xmath392 , the case of @xmath397 is similar .
let @xmath531 ; the elements @xmath59 , @xmath60 , @xmath79 and @xmath80 must be circularly ordered .
now , since @xmath61 is the unique element of @xmath532 , we know that @xmath533 .
similarly , @xmath534 .
but then @xmath535 , and the inequality is strict because @xmath536 .
this contradicts that @xmath87 , which is @xmath537 , is the @xmath538 minimal element of @xmath30 .
the aim of the next several propositions is to establish : [ winding ] let @xmath29 be a connected grassmann necklace .
let @xmath1 be weakly separated from @xmath29 , but not an element of @xmath29 .
then @xmath539 is in the interior of @xmath521 if and only if @xmath1 is in @xmath30 .
let @xmath540 be the parameterization of the simple closed curve @xmath521 , with @xmath541 , and @xmath542 for @xmath543 .
we consider the path @xmath544 in @xmath545 .
the point @xmath539 is inside @xmath521 if and only if @xmath544 has winding number @xmath546 ; if not , it has winding number @xmath547
. we will prove proposition [ winding ] by computing this winding number .
we first introduce some notation . by hypothesis , @xmath115 and @xmath1
are weakly separated .
let @xmath548 $ ] be the smallest cyclic interval containing all the elements of @xmath549 , and @xmath550 $ ] the smallest cyclic interval containing all the elements of @xmath551 .
since @xmath552 , these intervals are nonempty .
so @xmath402 , @xmath553 , @xmath554 and @xmath555 are circularly ordered , with @xmath556 and @xmath557 .
now , @xmath558 let @xmath559 be the centroid of @xmath560 , and @xmath561 the centroid of @xmath562 ; since @xmath115 and @xmath1 are weakly separated , @xmath563 .
so @xmath564 is parallel to @xmath565 .
[ lem : intervals ] there are circular intervals @xmath566 $ ] , and @xmath567 $ ] , with @xmath568 circularly ordered , so that @xmath548 $ ] and @xmath569 $ ] are contained in @xmath566 $ ] and @xmath550 $ ] and @xmath570 $ ] are contained in @xmath567 $ ] .
we have to consider four cases , depending on whether or not @xmath116 and @xmath571 are in @xmath1 .
let s consider the case that neither of them is . in this case ,
@xmath550 = [ c_{r+1 } , d_{r+1}]$ ] and we can take @xmath567 = [ c_r , d_r]$ ] .
the intervals @xmath548 $ ] and @xmath569 $ ] are disjoint from @xmath550 $ ] , and thus live in @xmath572 .
it is thus easy to see that there is a minimal subinterval of @xmath572 which contains @xmath548 $ ] and @xmath569 $ ] ; we take this interval to be @xmath566 $ ] .
the other three cases are similar to this one .
for @xmath372 a real number between @xmath573 and @xmath574 , define @xmath575 by linear interpolation between @xmath559 and @xmath576 .
similarly , define @xmath577 by linear interpolation between @xmath561 and @xmath578 .
for @xmath372 between @xmath573 and @xmath574 , the point @xmath575 is in the convex hull of @xmath579}$ ] and @xmath577 is in the convex hull of @xmath580}$ ] . because the intervals @xmath566 $ ] and @xmath567 $ ] are disjoint , we know that @xmath581 .
the vector @xmath544 is a positive scalar multiple of @xmath582 .
let @xmath583 be a convex simple closed curve through the points @xmath387 .
let @xmath584 and @xmath585 be the points where the line through @xmath575 and @xmath577 meets @xmath583 , with @xmath584 closer to @xmath575 .
so @xmath584 lies on the part of @xmath583 between @xmath586 and @xmath587 ; the point @xmath585 lies on the part of @xmath583 between @xmath588 and @xmath589 .
note that @xmath584 and @xmath585 are continuous functions of @xmath372 .
it is geometrically clear that the paths @xmath584 and @xmath585 , travelling around @xmath583 , have the same winding number as the path @xmath590 around @xmath547 . as observed above
, the vector @xmath590 is a positive multiple of @xmath544 .
thus , we are reduced to computing the winding number of @xmath584 around @xmath583 .
first , suppose that @xmath199 .
this means that , for every @xmath116 , we have @xmath591 ; since @xmath592 , this inequality is in fact strict .
the indices @xmath593 are circularly ordered ; the inequality between @xmath553 and @xmath554 are strict , while the others are weak .
so the interval @xmath566 $ ] can only contain @xmath116 at its left extreme .
for @xmath372 between @xmath573 and @xmath574 , the point @xmath584 lies in the interval @xmath566 $ ] , so @xmath584 has winding number @xmath546 , as desired . now
, suppose that @xmath594 .
then , for some @xmath595 $ ] , we have @xmath596 ; by hypothesis , @xmath597 .
so one of the following two cases must hold : 1 .
the minimal element of @xmath598 is less than the minimal element of @xmath599 or 2 .
the maximal element of @xmath599 is greater than the maximal element of @xmath598 .
here and in the rest of this proof , minimal " , maximal " , less than " and greater than " refer to the orders @xmath600 .
we discuss the former case ; the latter is similar .
we know that @xmath601 is minimal among the sets @xmath115 .
so , for every @xmath116 , the minimal element of @xmath551 must be less than the minimal element of @xmath549 , in the @xmath600 order .
so , for every @xmath116 , the interval @xmath548 $ ] does not contain @xmath602 .
so @xmath566 $ ] also does not contain @xmath602 , and we see that @xmath584 is not @xmath603 , for any @xmath372 .
so @xmath584 has winding number @xmath547 , as desired .
we now apply our constructions to the case that @xmath45 is a plabic graph .
[ thrm : dual ] let @xmath45 be a bipartite plabic graph whose strand permutation is connected and let @xmath304
. then @xmath386 is isomorphic to the dual complex to the planar graph @xmath45 .
also , @xmath604 fills the region surrounded by @xmath521 .
if @xmath45 is a reduced plabic graph which may not be bipartite , let @xmath289 be the graph obtained by contracting edges between vertices of the same color , as discussed in remark [ rem : contract ] .
then the faces of @xmath45 and @xmath289 are in bijection , and they have the same labels . letting @xmath14 be this common set of labels , it then follows from theorem [ thrm : dual ] that @xmath386 is dual to @xmath289 .
if @xmath45 is a bipartite plabic graph whose strand permutation is _ not connected _ , then @xmath386 is a union of @xmath605 s for various smaller collections @xmath83 , by proposition [ glueatpoint ] .
each of these collections comes from a connected plabic graph @xmath289 , to which theorem [ thrm : dual ] applies . by proposition [ prop : plagws ]
, @xmath14 is weakly separated , so it makes sense to define @xmath386 . by definition ,
the vertices of @xmath386 correspond to the faces of @xmath45 , with a separate vertex for each face of @xmath45 by proposition [ prop : partconverse ] . by propositions [ prop :
embed ] and [ winding ] , we know that @xmath386 embeds into @xmath385 , with the boundary vertices corresponding to @xmath29 .
let @xmath338 be the region in @xmath385 surrounded by the boundary of @xmath606 .
let @xmath607 be the dual to the planar graph @xmath45 .
so @xmath607 is a two - dimension cw - complex , homemorphic to a disc , with vertices labeled by @xmath14 , and with boundary labeled by @xmath29 .
we now describe a map @xmath608 of @xmath607 into @xmath385 : for a vertex @xmath288 of @xmath607 , if @xmath1 is the label of the dual face of @xmath45 , then @xmath609 is @xmath539 . for each edge @xmath357 of @xmath607 ,
connecting @xmath287 and @xmath288 , let @xmath610 be a line segment from @xmath611 to @xmath609 .
we now discuss faces of @xmath607 .
let @xmath300 be a face of @xmath607 , dual to a vertex @xmath288 of @xmath45 .
suppose that @xmath288 is black ; the case where @xmath288 is white is similar .
let the edges ending at @xmath288 cross the strands @xmath53 , @xmath54 ,
, @xmath55 .
so the faces bordering @xmath288 have labels of the form @xmath612 , @xmath613 ,
, @xmath614 for some @xmath395 .
in particular , we see that the corresponding vertices of @xmath607 are in @xmath397 .
the face @xmath397 is embedded in @xmath385 as a convex polygon .
moreover , the vertices of @xmath300 occur in cyclic order among those of @xmath397 .
so @xmath608 embeds @xmath615 as a convex polygon in @xmath385 , and we extend @xmath608 to take the interior of @xmath300 to the interior of that polygon . note that we have shown that , for every face @xmath300 of @xmath607 , the image @xmath616 lies within a face of @xmath386 of the same color .
the map @xmath608 takes the boundary of @xmath607 to @xmath617 . for every face @xmath300 of @xmath607 , the face @xmath616
is embedded in @xmath385 with the correct orientation . since @xmath607 is homeomorphic to a disc , these points force @xmath608 to be a homeomorphism onto @xmath338 .
every face of @xmath618 lies in a face of @xmath606 , so @xmath619 . by proposition
[ winding ] , we also have @xmath620 , so we conclude that @xmath621 .
we now must show that the two cw - structures on this subset of @xmath385 coincide .
let @xmath300 be any face of @xmath607 .
we consider the case that @xmath300 is black ; the white case is similar .
as shown above , there is a black face @xmath622 of @xmath386 which contains @xmath616 .
suppose , for the sake of contradiction , that some point of @xmath622 is not in @xmath623 .
since @xmath607 covers all of @xmath338 , there must be some face @xmath302 of @xmath607 , adjacent to @xmath300 , such that @xmath624 overlaps @xmath622 .
but @xmath302 , being a neighbor of @xmath300 , is white and thus lies in some @xmath625 .
so @xmath622 and @xmath625 overlap , a contradiction .
we conclude that every face of @xmath618 is also a face of @xmath606 .
since @xmath608 is injective , and has image @xmath338 , we see that every face of @xmath606 is precisely one face of @xmath618 .
we now have a bijection between the faces of @xmath386 and of @xmath607 . in both @xmath606 and @xmath618 ,
a face is the convex hull of its vertices , so the corresponding faces of @xmath386 and @xmath607 must have the same vertices and the same edges .
so @xmath626 is a cell - by - cell homeomorphism between the simplicial complexes @xmath386 and @xmath607 .
the most technical part of this paper is establishing that , when @xmath14 is a maximal weakly separated collection which is not yet known to be of the form @xmath301 , then @xmath604 fills the interior of the region surrounded by @xmath521 .
this task is accomplished in section [ sec : final ] .
we now show that @xmath301 is always maximal .
[ thrm : ismaximal ] let @xmath45 be a reduced plabic graph , with boundary @xmath29 . then @xmath627 is a maximal weakly separated collection in @xmath107 .
we first reduce to the case that @xmath29 is connected .
if not , let @xmath628 .
we resume the notations of section [ connected ] .
the face of @xmath45 labeled @xmath91 touches the boundary of the disc at two points ; both between boundary vertices @xmath61 and @xmath629 and between boundary vertices @xmath111 and @xmath630 .
so this face disconnects the graph @xmath45 into two graphs , with vertices in @xmath631 and the other with vertices in @xmath229 .
call these plabic graphs @xmath632 and @xmath633 .
it is an easy exercise that each @xmath634 is a reduced plabic graph , with boundary @xmath635 . by induction on @xmath18 , we know that @xmath636 is a maximal weakly separated collection in @xmath637 .
let @xmath1 in @xmath107 be weakly separated from @xmath301 .
since @xmath638 is weakly separated , by proposition [ splittingsummary ] , @xmath1 must either be of the form @xmath274 or @xmath275 , where @xmath227 is weakly separated from @xmath639 and in @xmath640 .
but , by the maximality of @xmath639 , this implies that @xmath227 is in @xmath639 .
then @xmath1 is the label of the corresponding face of @xmath301 .
we have shown that any @xmath1 in @xmath107 which is weakly separated from @xmath301 must lie in @xmath301 , so @xmath301 is maximal in @xmath107 .
we now assume that @xmath29 is connected .
let @xmath1 in @xmath107 be weakly separated from @xmath301 and assume , for the sake of contradiction , that @xmath641 .
we showed above that @xmath604 fills the interior of the region surrounded by @xmath521 . by proposition [ winding ] , @xmath539 is in this region , so @xmath539 lands on some point of @xmath604 .
in particular , @xmath31 is not injective on @xmath642 , contradicting the assumption that @xmath1 is weakly separated from @xmath14 .
this contradiction shows that @xmath301 is maximal in @xmath107 .
our aim in this section is to prove the following technical lemma . the reader may wish to skip this proof on first reading , and go on to see how it is used in the proof of theorem [ thm : wsispg ] .
[ lem : key ] let @xmath14 be a weakly separated collection of @xmath23}{k}$ ] .
let @xmath643}{k-1}$ ] and @xmath644 $ ] be such that @xmath645 and @xmath646 are in @xmath14 . if @xmath647 is not the interval @xmath477 , then there is a set of one of the following two types which is weakly separated from @xmath14 : 1 .
@xmath648 with @xmath649 or 2 .
@xmath650 with @xmath651 .
note that this lemma is taken to itself under the symmetry which replaces every subset of @xmath2 $ ] by its complement and switches @xmath652 and @xmath653 .
we call this symmetry * _ dualization_*. the condition that @xmath654 means that either there exists some @xmath649 or there exists some @xmath651 ( or both ) .
consider under what circumstances a set of the form @xmath655 might not be weakly separated from @xmath14 . in order for this to happen , there must be some @xmath1 in @xmath14 such that @xmath656 .
the next lemma studies the properties of such a @xmath1 . [ witnessdescription ]
let @xmath647 , @xmath652 and @xmath653 be as above , and let @xmath649 .
suppose that @xmath657 , @xmath658 but @xmath659 .
then there are elements @xmath28 and @xmath116 such that : 1 .
@xmath28 and @xmath116 are in @xmath660 2 .
@xmath661 3 .
4 . @xmath663 .
@xmath664 note that @xmath1 is a vertex of @xmath386 , and @xmath665 is contained in a face of @xmath386 .
( it may or may not be an edge . ) then , by lemma [ writethis ] , the vectors @xmath666 and @xmath667 are weakly separated . by hypothesis , @xmath666 and @xmath668
are not weakly separated .
the only coordinates to change between @xmath669 and @xmath670 are in positions @xmath652 , @xmath653 and @xmath59 .
coordinates @xmath652 and @xmath653 increase by @xmath671 from half integers to integers .
in particular , they can not change from nonnegative to negative or nonpositive to positive .
so , the failure of weak separation between @xmath1 and @xmath672 must be attributable to coordinate @xmath59 .
we must have @xmath673 , and we must have @xmath664 for some @xmath28 and @xmath674 such that @xmath662 . moreover , we choose to take the interval @xmath675 to be of minimal length , subject to the conditions that @xmath664 and @xmath676 and @xmath674 .
this ensures that @xmath677 .
we are left to check that @xmath28 and @xmath116 are in @xmath678 $ ] .
we check that @xmath28 is ; the case of @xmath116 is similar .
suppose , instead , that @xmath679 .
if @xmath680 then @xmath681 would be closer to @xmath59 then @xmath682 , contradicting our minimal choice of @xmath675 .
so @xmath683 .
looking at @xmath684 , we see that @xmath1 and @xmath685 are not weakly separated , a contradiction .
we define @xmath686 to be a * _ witness against @xmath655 _ * if @xmath273 and the numbered conditions of lemma [ witnessdescription ] hold .
[ overlappingwitnesses ] suppose that @xmath687 is nonempty , and no @xmath655 is weakly separated from @xmath14 .
then there is a sequence of triples , @xmath688 , ,
@xmath689 , each of which obeys , 1 .
@xmath690 2 .
@xmath691 and @xmath692 are in @xmath693 3 .
@xmath694 4 . @xmath695 .
5 . @xmath696 . and such that @xmath697 and @xmath698 and @xmath699 .
consider any @xmath649 .
since @xmath655 is not weakly separated from @xmath14 , there is some @xmath1 in @xmath655 which is not weakly separated from @xmath655 . since @xmath700 and @xmath685 are in @xmath14 , we know that @xmath1 is weakly separated from @xmath700 and @xmath685 .
so we can complete @xmath1 to a witness , @xmath686 , against @xmath655 .
this @xmath686 will obey the numbered conditions above . choose a minimal collection @xmath701 , , @xmath689 , obeying the numbered conditions , so that every @xmath649 lies in some @xmath702 .
( since there is some @xmath649 , this minimal collection is nonempty . )
we can not have @xmath703 for any @xmath192 , as otherwise we could delete @xmath704 and have a smaller collection .
thus , we may reorder our collection so that @xmath705 and @xmath698 .
if @xmath706 , then @xmath707 is an element of @xmath687 and @xmath707 does not lie in any @xmath702 , a contradiction .
similarly , @xmath708 .
so @xmath709 and @xmath710 .
finally , suppose that @xmath711 .
then @xmath712 is an element of @xmath687 which is not in any @xmath702 , a contradiction .
we now assume that @xmath687 is nonempty and that no @xmath655 is weakly separated from @xmath14 . for clarity , we will continue to state these assumptions explicitly in all lemmas that rely on them .
we fix , once and for all , a sequence of triples @xmath713 as in lemma [ overlappingwitnesses ] .
we introduce a total order @xmath714 on @xmath23}{k}$ ] : given @xmath0 and @xmath1 in @xmath23}{k}$ ] , first compare @xmath715 and @xmath716 . if @xmath717 , set @xmath718 . if this comparison is inconclusive and @xmath719 , set @xmath718 .
if both of these comparisons are inconclusive , and @xmath720 is lexicographically before @xmath721 , set @xmath718 ; if that too is inconclusive , and @xmath722 is lexicographically after @xmath723 , set @xmath718 .
( here we use the orderings @xmath724 and @xmath725 on @xmath477 and @xmath484 respectively . ) finally , if all of these are inconclusive , then either @xmath726 or @xmath727 for some @xmath64 ; set @xmath728 .
this order is constructed to obey the dual lemmas : [ lw - works ] suppose that @xmath0 and @xmath1 are weakly separated .
suppose that there are @xmath729 in @xmath678 $ ] such that @xmath59 and @xmath730 and @xmath731
. then @xmath718 .
[ lw - worksdual ] suppose that @xmath0 and @xmath1 are weakly separated .
suppose that there are @xmath729 in @xmath732 $ ] such that @xmath59 and @xmath733 and @xmath734 .
then @xmath718 .
we prove lemma [ lw - works ] and leave it to the reader to dualize our argument to prove lemma [ lw - worksdual ] .
since @xmath0 and @xmath1 are weakly separated , there can be no element of @xmath732 $ ] which is in @xmath8 .
if there is any element of @xmath4 in @xmath732 $ ] , then @xmath735 , so we have @xmath718 as desired .
if there is no element of @xmath4 in @xmath732 $ ] , then @xmath736 and @xmath737 , so we move to the third and fourth prongs of the test .
we have @xmath738 .
we know that @xmath739 , and , because @xmath0 and @xmath1 are weakly separated , there can be no element of @xmath740 which is @xmath741-prior to @xmath59 .
so @xmath718 in this case as well .
we make the following definition : for @xmath742 , we say that @xmath743 is an * _ @xmath356-snake _ * if 1 . all of the integers @xmath744 , @xmath745 ,
, @xmath691 , @xmath692 , , @xmath746 , @xmath747 are in @xmath743 and 2 .
@xmath748 is nonempty .
note that @xmath743 is always an @xmath749-snake .
the following lemma allows us to construct snakes .
[ snakeexists ] assume that @xmath687 is nonempty and that no @xmath655 is weakly separated from @xmath14 . for any indices @xmath61 and
@xmath111 with @xmath750 , let @xmath743 be the @xmath714-minimal witness among @xmath751 , @xmath752 , , @xmath753 .
then @xmath743 is an @xmath356-snake . in particular , for any @xmath356 , there is an @xmath356-snake .
our proof is by induction on @xmath754 . when @xmath178 , the result is clear .
we now consider the case that @xmath755 .
we first establish that @xmath744 , @xmath745 ,
, @xmath691 , @xmath692 , , @xmath746 , and @xmath747 are in @xmath743 .
suppose , for the sake of contradiction , that @xmath692 , @xmath756 ,
, @xmath757 are in @xmath743 but @xmath758 is not . by induction ,
there is a @xmath759 snake @xmath760 .
then @xmath761 and @xmath758 are in @xmath760 and not in @xmath743 , while @xmath762 is in @xmath763 and not @xmath760 by lemma [ lw - works ] .
this means that @xmath764 , a contradiction .
we conclude that @xmath692 ,
, @xmath746 , and @xmath747 are in @xmath743 ; similarly , @xmath744 , , @xmath765 and @xmath691 are in @xmath743 .
we now establish that @xmath748 is nonempty . we know
that @xmath766 is nonempty ; let @xmath767 .
suppose , for the sake of contradiction , that @xmath748 is empty , and hence @xmath768 for some @xmath769 .
( the case where @xmath770 is similar . ) by induction , there is an @xmath759-snake @xmath760 .
let @xmath771 be an element of @xmath772 .
@xmath700 and @xmath760 are weakly separated .
considering @xmath773 , we see that @xmath774 .
but then @xmath761 and @xmath357 are in @xmath760 and not @xmath743 , while @xmath762 and @xmath771 are in @xmath743 and not @xmath760 . by our choice of @xmath769 such that @xmath768 , we know that @xmath775so @xmath760 and @xmath743 are not weakly separated , a contradiction . as a corollary
, we remove the need to assume that @xmath687 is nonempty .
assume that @xmath14 is not weakly separated from any @xmath655 nor any @xmath776 , and recall the standing hypothesis that @xmath647 is not the interval @xmath477 .
then both @xmath687 and @xmath777 are nonempty . by our standing hypotheses ,
either @xmath687 or @xmath777 is nonempty . without loss of generality ,
assume it is the former so the @xmath713 are defined .
lemma [ snakeexists ] applies , and we know that there is an @xmath778-snake @xmath743 . for this @xmath743 ,
we know that @xmath779 is nonempty . since @xmath780 , we conclude that @xmath777 is also nonempty .
we are now entering into the final stages of the proof of lemma [ lem : key ] .
assume , for the sake of contradiction , that @xmath14 is not weakly separated from any @xmath655 nor any @xmath776 .
so we have a sequence of witnesses @xmath713 , and a dual sequence @xmath781 .
we point out that @xmath782 and @xmath783 are in @xmath732 $ ] ; we leave it to the reader to dualize the other parts of the definition of a witness .
let @xmath784 and @xmath785 be the @xmath714-minimal elements among the sequences @xmath751 and @xmath786 .
[ musarewide ] @xmath784 contains both @xmath652 and @xmath653 , and @xmath787 is nonempty .
@xmath785 contains neither @xmath653 nor @xmath652 and @xmath788 is nonempty .
the first statement is lemma [ snakeexists ] applied to @xmath789 .
the second statement is the dual of the first .
let @xmath790 . by lemma [ musarewide ] ,
@xmath791 is nonempty ; let @xmath792 . by definition ,
@xmath649 , so there is some @xmath793 which is a witness against @xmath655 . in particular , @xmath794 contains @xmath59 , and is thus nonempty .
let @xmath795 be @xmath714-minimal such that @xmath796 is nonempty . with the above definition , @xmath797 . for the sake of contradiction ,
assume that @xmath798 .
let @xmath799 be in @xmath796 .
then @xmath799 is in @xmath785 and not in @xmath800 . by the contrapositive of lemma [ lw - works ] ,
either @xmath785 contains every element of @xmath801 , or @xmath785 contains every element of @xmath802 $ ] . without loss of generality ,
assume that it is the latter . since @xmath803 ( lemma [ musarewide ] )
, we deduce that @xmath804 .
now , @xmath805 .
let @xmath111 be such that @xmath806 , @xmath807 ,
, @xmath746 are in @xmath800 and @xmath747 is not .
let @xmath743 be the @xmath714-minimal element of @xmath800 , @xmath808 ,
, @xmath809 . by lemma [ snakeexists ]
, @xmath743 is a @xmath810-snake . since @xmath811 , we know that @xmath812 and thus we have the strict inequality @xmath813 . since @xmath814 , we have @xmath815 . since @xmath816 $ ] , we have @xmath817 . also , one of the conditions on the witness @xmath818 is that @xmath819 .
so @xmath820 is in @xmath791 .
but @xmath820 is in @xmath702 and we saw above that @xmath813 .
this contradicts the minimality of @xmath800 .
we have now shown that there is some @xmath800 with @xmath797 . by the minimality of @xmath784 , we have @xmath821 , so @xmath822 .
but the dual argument shows that @xmath823 .
we have reached a contradiction , and lemma [ lem : key ] is proved .
in this section , we prove theorem [ thm : gsconjecture ] . more precisely , we prove : [ thm : wsisfgpos ] let @xmath29 be a grassmann necklace , @xmath30 the associated positroid .
let @xmath14 be a subset of @xmath23}{k}$ ] .
then @xmath14 is a maximal weakly separated collection in @xmath30 if and only if @xmath14 is of the form @xmath301 for a plabic graph @xmath45 whose decorated permutation , @xmath297 corresponds to the grassmann necklace @xmath29 . see section [ sec : grneck ] to review the definitions of positroids , grassmann necklaces and decorated permutations . by theorem [ thrm : ismaximal ] ,
if @xmath14 is of the form @xmath301 , then @xmath14 is a maximal weakly separated collection in @xmath30 .
so our goal is to prove the converse .
suppose that @xmath14 is a maximal weakly separated collection in @xmath30 .
we first reduce to the case that @xmath29 is connected .
suppose that @xmath191 .
we reuse the notations @xmath631 , @xmath229 and so forth from section [ connected ] .
so , by proposition [ splittingsummary ] , @xmath268 and @xmath269 are maximal weakly separated collections in @xmath207 and @xmath208 . by induction
, there are reduced plabic graphs @xmath632 and @xmath633 , with @xmath824 , and with boundary regions labeled @xmath631 and @xmath229 .
gluing these graphs along these boundary regions , we have a reduced plabic graph @xmath45 with @xmath825 .
thus , we may now assume that @xmath29 is connected .
form the @xmath826-complex @xmath386 . by proposition [ prop :
embed ] , the map @xmath31 embeds @xmath386 into @xmath385 . by proposition [ winding ] , for every @xmath827 , the point @xmath539 is inside the curve @xmath521 . since these are the vertices of @xmath386 , we see that @xmath386 lies within this curve .
thus , @xmath386 is a finite polyhedral complex in @xmath385 , with outer boundary @xmath521 .
we will eventually take @xmath45 to be the dual graph to @xmath386 . in order to do this ,
we need the following proposition : [ noholes ] if @xmath14 is a maximal weakly separated collection for the connected positroid @xmath30 , then @xmath386 fills the entire interior of the curve @xmath521 .
[ incompleteexample ] consider the weakly separated collection @xmath828 for the grassmann necklace @xmath829 .
this collection only has @xmath830 elements , not @xmath831 which a maximal collection would have .
figure [ incomplete ] shows the corresponding @xmath386 , embedded in @xmath385 .
the grey region , labeled @xmath832 , is not in @xmath386 , but note that all four edges of its boundary are in @xmath386 . in the vocabulary of the proof of proposition [ noholes ] , we take @xmath833 , @xmath834 and @xmath835 .
our strategy is to use lemma [ lem : key ] to show that @xmath14 must be weakly separated from one of @xmath836 , which is @xmath837 , or @xmath838 , which is @xmath839 .
( in fact , it is weakly separated from both . )
let @xmath840 denote the interior of @xmath521 .
suppose , to the contrary , that @xmath832 is a connected component of @xmath841 .
let @xmath842 be an edge in @xmath843 , with @xmath832 on the lefthand side as we look from @xmath844 to @xmath845 .
note that @xmath665 is not of the form @xmath846 as , if it were , then @xmath832 would be on the outside of @xmath521 .
in particular , @xmath647 is not the interval @xmath484 . by lemma [ lem : key ]
, there is a set @xmath1 , either of the form @xmath672 , with @xmath649 , or the form @xmath847 , with @xmath651 , such that @xmath1 is weakly separated from @xmath14 .
the conditions on @xmath59 and @xmath60 imply that the triangle @xmath848 lies on the @xmath832 side of @xmath842 .
( the reader is invited to check this in example [ incompleteexample ] . )
if @xmath199 then @xmath1 must be in @xmath14 , by the maximality of @xmath14 .
but then the triangle @xmath848 is part of @xmath386 , and lies on the @xmath832-side of @xmath842 , contradicting the supposition that @xmath842 is a boundary edge of @xmath832 .
so @xmath849 .
( at this point , we are already done if @xmath30 is the uniform matroid . )
we have shown that @xmath849 . by proposition [ winding ] , @xmath539 lies outside of @xmath521 . in particular
, the triangle @xmath848 must cross some boundary edge @xmath528 .
let s say that segments @xmath850 and @xmath528 cross ( the case of @xmath685 is practically identical ) .
note that @xmath851 and @xmath852 are both subsets of cliques in the weakly separated collection @xmath853 .
it almost contradicts proposition [ prop : embed ] for these two segments to cross .
the only subtlety is that @xmath851 and @xmath852 might be contained in the same clique of @xmath853 , with @xmath854 occurring in that circular order around the corresponding face of @xmath855 .
but then the triangle @xmath856 is contained in @xmath386 , and lies on the @xmath832-side of @xmath665 , contradicting that @xmath665 is supposed to be on the boundary of @xmath832 .
so , @xmath386 is a disc , whose two faces are naturally colored black and white .
let @xmath45 be the dual graph , so the faces of @xmath45 are labeled by @xmath14 , and the boundary faces by @xmath29 .
there are @xmath18 vertices on the boundary of @xmath45 , labeled by @xmath2 $ ] , with vertex @xmath61 between faces @xmath91 and @xmath87 .
we now show that @xmath45 is reduced .
be warned that , at this point , we do not know that the labels of the faces of @xmath45 inherited from @xmath386 are the labels @xmath301 .
let @xmath583 be a strand which separates @xmath672 and @xmath700 , for some @xmath652 , and then goes on to separate @xmath672 from some @xmath685 .
as we travel along @xmath583 , the faces on the left are of the form @xmath857 for some sequence @xmath858 , with @xmath859 , and the faces on the right are of the form @xmath860 .
we have @xmath861 due to the definition of cliques in plabic tilings .
therefore , @xmath583 can not loop or cross itself ; it must travel from the boundary of @xmath840 to the boundary of @xmath840 .
in particular , we see that there are only @xmath18 strands in total , one ending between each pair of boundary faces of @xmath45 .
the strand @xmath583 must start separating @xmath862 from @xmath863 , and end separating @xmath864 from @xmath865 .
that is because these are the only pairs of adjacent faces at the boundary of @xmath45 which differ by an @xmath59 .
thus , the strand we have called @xmath583 is the strand labeled by @xmath59 in our standard way of labeling the strands of a plabic graph .
now fix two strands @xmath291 and @xmath292 which intersect at least twice , labeled @xmath59 and @xmath60 .
we want to show that the intersections occur in reverse order along @xmath291 and @xmath292 . at one crossing , let the strands separate faces @xmath866 and @xmath867 ; at the other crossing let the adjacent faces be @xmath868 and @xmath869 . from the fact that @xmath870 and @xmath871 , we get that one of the following holds : * @xmath872 and @xmath873 or * @xmath874 and @xmath875 .
assuming the first case holds , we get @xmath876 and @xmath877 .
so the intersections occur in opposite order along @xmath291 and @xmath292 .
we also conclude that the intersections occur in opposite order in the second case .
we have now checked that strands do not cross themselves , do not form closed loops , and that the intersections along any pair of strands occur in opposite order along the two strands .
so @xmath45 is reduced .
finally , we must check that the labels coming from @xmath386 are the same as the labels @xmath301 . in this paragraph , when we refer to the label of a face of @xmath45 , we mean the label @xmath0 of the dual vertex of @xmath386 .
consider a strand @xmath583 , with label @xmath59 .
we must check that the faces to the left of @xmath583 all contain @xmath59 , and those to the right do not .
the faces immediately to the left of @xmath583 all contain @xmath59 , and those immediately to the right do not .
let @xmath878 be the union of all faces that contain @xmath59 , so @xmath583 is part of the boundary of @xmath878 , as is part of @xmath879 .
we claim that , in fact , this is the entire boundary of @xmath878 .
if there were any other edge @xmath357 in the boundary of @xmath878 , since @xmath45 is a disc with boundary @xmath879 , there would have to be a face on the other side of @xmath357 from @xmath878 , and the label of this face would not contain @xmath59 .
but then one of the strands passing through this edge would be @xmath583 .
similarly , the boundary of @xmath880 is also made up of @xmath583 and a piece of @xmath879 .
it is now topologically clear that @xmath878 is the part of @xmath840 on one side of @xmath583 , as desired .
so starting from a maximal weakly separated collection @xmath14 for @xmath30 , we have obtained a reduced plabic graph @xmath45 with @xmath383 .
this completes the proof of theorem [ thm : wsisfgpos ] . .
we have now proved theorem [ thm : wsisfgpos ] . by theorem [ thm : p ] , we now know that every maximal weakly separated collection of @xmath107 has cardinality @xmath881 and any two maximal weakly separated collections of @xmath30 are mutation equivalent .
this is theorem [ thm : gsconjecture ] and our work is complete .
before we end , we state a simple result that follows from theorem [ thm : wsisfgpos ] .
[ cor : wsalign ] let @xmath29 be a grassmann necklace , @xmath169 the associated decorated permutation , and @xmath30 the associated positroid .
we define @xmath882 to be the union of @xmath354 for all reduced plabic graphs @xmath45 of @xmath30 .
then @xmath199 is in @xmath882 if and only if it obeys the following condition : for any alignment @xmath158 in @xmath31 , if @xmath883 then we have @xmath884 .
one direction is obvious : for any @xmath368 for some plabic graph @xmath45 of @xmath30 , it should meet the condition due to the way the faces are labeled in plabic graphs .
so we only need to check the other direction .
pick any @xmath199 that satisfy the condition .
assume for the sake of contradiction that @xmath1 is not weakly separated with some @xmath91 .
this means that there is some @xmath885 and @xmath886 such that @xmath887 is circularly ordered .
the condition @xmath888 tells us that @xmath889 , while the condition @xmath890 tells us that @xmath891 .
so @xmath892 are circularly ordered and @xmath893 forms an alignment of @xmath31 .
then @xmath1 should contain both @xmath17 and @xmath372 due to the condition , but this contradicts @xmath894 .
so @xmath1 is weakly separated from all @xmath91 s , and @xmath895 is a weakly separated collection
. then theorem [ thm : wsisfgpos ] tells us that there is some @xmath354 that contains @xmath1 .
in this section , we state leclerc and zelevinsky s conjectures on @xmath35-chamber sets and explain how they follow from theorem [ thm : gsconjecture ] . given two sets of integers , @xmath0 and @xmath1
, we write @xmath896 if @xmath755 for all @xmath897 and all @xmath884
. we will write @xmath898 for @xmath899 with respect to the shifted order @xmath99 on @xmath2 $ ]
. the following definition will only be necessary in this section ; it is an extension of the definition of weak separation to sets of unequal cardinality .
let @xmath35 be a permutation in the symmetric group @xmath351 .
the @xmath35-chamber set , @xmath913 , is defined to be the collection of those subsets @xmath0 of @xmath914 $ ] that satisfy the following condition : for every pair @xmath915 with @xmath916 , if @xmath917 then @xmath918 .
the attentive reader will notice that our definition of a @xmath35-chamber set differs from that in @xcite by switching the directions of all inequalities .
thus , the @xmath35-chamber set in our notation is the image of leclerc and zelevinsky s @xmath919-chamber set under the map @xmath920 .
this flip has to be included somewhere , due to an incompatibility in the sign conventions of @xcite and @xcite .
a weakly separated collection @xmath14 of @xmath913 is a collection of subsets in @xmath913 such that the elements of @xmath14 are pairwise weakly separated ( in the lz sense ) .
a maximal weakly separated collection of @xmath913 is a weakly separated collection of @xmath913 that is maximal under inclusion .
let @xmath931 and @xmath932 . the padding allows us to reduce weak separation in the sense of leclerc and zelevinsky to the notion of weak separation of elements in @xmath23}{k}$ ] which we use in the rest of the paper . specifically , we have : if @xmath19 then @xmath935 if and only if @xmath936 if and only if @xmath934
otherwise , we may assume that @xmath937
. then @xmath938 .
note that @xmath8 is contained in @xmath914 $ ] , as is @xmath939 .
so @xmath925 is weakly separated from @xmath940 if and only if there is no element of @xmath8 which lies between two elements of @xmath4 in the standard linear order on @xmath914 $ ] .
this precisely means that we can partition @xmath8 as @xmath941 so that @xmath942 .
next , we explain the relation between positroids and @xmath35-chamber sets . for a permutation @xmath943 ,
let @xmath944 be the grassmann necklace corresponding to the decorated permutation @xmath945.\ ] ] ( the permutation @xmath946 has no fixed points , so we do not need to describe a coloring . )
let us compute the number of alignments of @xmath946 .
all @xmath952 pairs @xmath158 with @xmath953 $ ] form an alignment .
a pair @xmath954 with @xmath953 $ ] forms an alignment if and only if @xmath356 is not an inversion of @xmath955 , that is , @xmath755 and @xmath956
. there are @xmath957 alignments of this type .
finally , a pair @xmath958 with @xmath953 $ ] never forms an alignment . in total
, we get @xmath959 now let s check the second arugment . any pair @xmath960 with @xmath916 translates into an alignment @xmath961 of @xmath946 .
so @xmath951 implies @xmath950 .
the alignments @xmath961 of @xmath946 such that @xmath962 $ ] translates into the restriction that any @xmath963 is obtained by padding some subset of @xmath914 $ ] .
and as we have checked above , a pair @xmath958 with @xmath953 $ ] never forms an alignment .
so @xmath950 implies @xmath951 .
|
leclerc and zelevinsky described quasicommuting families of quantum minors in terms of a certain combinatorial condition , called _ weak separation .
_ they conjectured that all maximal by inclusion weakly separated collections of minors have the same cardinality , and that they can be related to each other by a sequence of mutations . on the other hand
, postnikov studied _ total positivity _ on the grassmannian .
he described a stratification of the totally nonnegative grassmannian into _ positroid strata , _ and constructed their parametrization using _ plabic graphs .
_ in this paper we link the study of weak separation to plabeic graphs .
we extend the notion of weak separation to positroids .
we generalize the conjectures of leclerc and zelevinsky , and related ones of scott , and prove them . we show that the maximal weakly separated collections in a positroid are in bijective correspondence with the plabic graphs .
this correspondence allows us to use the combinatorial techniques of positroids and plabic graphs to prove the ( generalized ) purity and mutation connectedness conjectures .
|
the measurement of the fluctuations in the kaon to pion ratio by the na49 collaboration was the first event - by - event analysis in nucleus - nucleus collisions @xcite .
it was suggested that this ratio might allow to distinguish events with enhanced strangeness production attributed to the qgp phase @xcite .
nowadays , the excitation function for this observable is available in a wide range of energies : from the na49 collaboration @xcite in pb+pb collisions at the cern sps and from the star collaboration @xcite in au+au collisions at rhic .
results from na49 show an enhancement of fluctuations in the kaon to pion multiplicity ratio for low energies which may be a signal of a deconfinement phase transition . on the other hand
there is no such an enhancement for the proton to pion ratio fluctuations .
this proceeding presents the results of a systematic study of @xmath0 , @xmath1 and @xmath2 ratio fluctuations ( @xmath3 , @xmath4 , and @xmath5 means @xmath6 ) based on the hsd transport model @xcite . for more details we refer readers to @xcite .
event - by - event fluctuations for charged hadron multiplicities also have been studied in the hsd transport approach @xcite .
let s introduce some notations .
the deviation @xmath7 from the average number @xmath8 of the particle species @xmath9 is defined by @xmath10 , while the covariance for species @xmath9 and @xmath11 is : @xmath12 the scaled variance and the correlation coefficient @xmath13^{1/2}}~. \end{aligned}\ ] ] the fluctuations of the ratio @xmath14 will be characterized by @xmath15 @xcite , which to the second order in @xmath16 and @xmath17 can be written as : @xmath18^{1/2}. \end{aligned}\ ] ]
the experimental data for @xmath19 fluctuations are usually presented in terms of the so called dynamical fluctuations @xcite
@xmath20 where @xmath15 is defined by ( [ sigma ] ) , and @xmath21 corresponds to the following mixed events procedure .
one takes a large number of nucleus - nucleus collision events , and measures the numbers of @xmath22 and @xmath23 in each event .
then all @xmath9 and @xmath11 particles from all events are combined into one _
set_. a construction of _ mixed events _ is done like the following : one fixes a random number @xmath24 according to the experimental probability distribution @xmath25 , takes randomly @xmath26 particles ( @xmath9 and/or @xmath11 ) from the _ whole set _ , fixes the values of @xmath22 and @xmath23 , and returns these @xmath26 particles into the _ set_. this is the mixed event number one .
then one constructs event number 2 , number 3 , etc .
note that the number of events is much larger than the number of hadrons , @xmath26 , in any single event .
therefore , the probabilities @xmath27 and @xmath28 to take the @xmath9 and @xmath11 species from the whole _
set _ can be considered as constant values during the event construction .
another consequence of a large number of events is the fact that all @xmath9 and @xmath11 particles in any constructed _ mixed event _ most probably belong to different _ physical events _ of nucleus - nucleus collisions .
therefore , the correlations between @xmath23 and @xmath22 numbers in a physical event are expected to be destroyed in a mixed event .
this is the main purpose of the mixed events construction .
calculating the @xmath19 fluctuations for mixed events one gets like in @xcite : @xmath29 the mixed event procedure gives the same @xmath15 for @xmath19 fluctuations as in the gce formulation for an ideal boltzmann gas , i.e. @xmath30 and @xmath31 in ( [ sigma ] ) .
note that non - trivial fluctuations of @xmath22 and @xmath23 ( @xmath32 ) as well as non - zero correlations @xmath33 may exist in the mixed events procedure , but they are canceled out exactly in @xmath34 .
the hsd results as well as statistical model calculations for ratio fluctuations in full acceptance are presented in @xcite a comparison of the sm results for fluctuations in different ensembles with the data looks problematic at present due to the difficulties in implementing the experimental acceptance and centrality selection which , however , can be taken into account in the transport approach . in order to compare the hsd calculations with the measured data ,
the experimental cuts are applied for the simulated set of hsd events . in fig .
[ fig : fig5 ] the hsd results of @xmath35 for the @xmath0 , @xmath2 and @xmath1 ratios are shown in comparison with the experimental data by the na49 collaboration at the sps @xcite and the preliminary data of the star collaboration at rhic @xcite .
the available results of urqmd calculations ( from refs .
@xcite ) are also shown by the dashed lines . -8pt
-8pt for the sps energies we use the na49 acceptance tables from ref . @xcite . for the rhic energies we use the following cuts
: in pseudorapidity , @xmath36 , and in transverse momentum , @xmath37 gev / c for kaons and pions and @xmath38 gev / c for protons @xcite .
note , that hsd results presented in fig .
[ fig : fig5 ] correspond to the centrality selection as in the experiment : the na49 data correspond to the 3.5% most central collisions selected via veto calorimeter , whereas in the star experiment the 5% most central events with the highest multiplicities in the pseudorapidity range @xmath39 have been selected .
one sees that the urqmd model gives practically a constant @xmath40 , which is by about @xmath41 smaller than the results from hsd at the lowest sps energy .
this difference between the two transport models may be probably attributed to different realizations of the string and resonance dynamics in hsd and urqmd : in urqmd the strings decay first to heavy baryonic and mesonic resonances which only later on decay to ` light ' hadrons such as kaons and pions . in hsd
the strings dominantly decay directly to ` light ' hadrons ( from the pseudoscalar meson octet ) or the vector mesons @xmath42 , @xmath43 and @xmath44 ( or the baryon octet and decouplet in case of baryon number @xmath45 ) .
such a ` non - equilibrated ' string dynamics may lead to stronger fluctuations of the @xmath0 ratio . at the sps energies the hsd simulations lead to negative values of @xmath35 for the proton to pion ratio .
this is in agreement with the na49 data in pb+pb collisions . on the other hand hsd
gives large positive values of @xmath46 at rhic energies which strongly overestimate the preliminary star data for au+au collisions @xcite .
for @xmath47 only preliminary star data in au+au collisions are available @xcite which demonstrate a qualitative agreement with the hsd results ( fig .
[ fig : fig5 ] ) .
the hsd results for @xmath47 show a weak energy dependence in both sps and rhic energy regions .
a peculiar feature is , however , a strong ` jump ' between the sps and rhic values , seen in the middle panel of fig .
[ fig : fig5 ] , in the hsd calculations which is caused by the different acceptances in the sps and rhic measurements .
the influence of the experimental acceptance is clearly seen at 160 a gev where a switch from the na49 to the star acceptance leads to the jump in @xmath47 by 3% - middle panel of fig .
[ fig : fig5 ] . on the other hand , our calculations for pb+pb ( 3.5% central ) and for au+au ( 5% central ) collisions - performed within the na49 acceptance for both cases at 160
a gev - shows a very week sensitivity of @xmath47 on the actual choice of the collision system and centrality cf . the coincident open circle and triangle at 160 a gev in the middle panel of fig .
[ fig : fig5 ] .
the event - by - event multiplicity fluctuations of @xmath0 , @xmath1 and @xmath2 ratio fluctuations in central au+au ( or pb+pb ) collisions from low sps up to top rhic energies have been studied within the hsd transport approach .
it has been found that the hsd model can qualitatively reproduce the measured excitation function for the @xmath0 ratio fluctuations in central au+au ( or pb+pb ) collisions from low sps up to top rhic energies .
accounting for the experimental acceptance as well as the centrality selection has a relatively small influence on @xmath35 and does not change the shape of the @xmath35 excitation function .
the hsd results for @xmath46 appear to be close to the na49 data at the sps .
the data for @xmath47 in pb+pb collisions at the sps energies will be available soon and allow for further insight .
a comparison of the hsd results with preliminary star data in au+au collisions at rhic energies are not fully conclusive : @xmath35 from hsd calculations is approximately in agreement with data @xcite for the kaon to proton ratio , but overestimate the experimental results @xcite for the proton to pion ratio .
new data on event - by - event fluctuations in au+au at rhic energies will help to clarify the situation .
this work was supported by the helmholtz international center for fair within the framework of the loewe program ( landesoffensive zur entwicklung wissenschaftlich - konomischer exzellenz ) launched by the state of hesse .
10 na49 , s. v. afanasev _
et al . _ ,
rev . lett . * 86 * , 1965 ( 2001 ) , hep - ex/0009053 .
s. jeon and v. koch , phys .
lett . * 83 * , 5435 ( 1999 ) , nucl - th/9906074 .
na49 , c. alt _ et al .
_ , ( 2008 ) , 0808.1237 .
star , s. das , j. phys .
* g32 * , s541 ( 2006 ) . b. i. abelev , ( 2009 ) , 0901.1795 . w. ehehalt and w. cassing , nucl .
a * 602 * , 449 ( 1996 ) ; e.l .
bratkovskaya and w. cassing , nucl .
a * 619 * 413 ( 1997 ) ; m. i. gorenstein , m. hauer , v. p. konchakovski , and e. l. bratkovskaya , phys . rev . * c79 * , 024907 ( 2009 ) , 0811.3089 . v. p. konchakovski , m. hauer , m. i. gorenstein , and e. l. bratkovskaya , j. phys . *
g36 * , 125106 ( 2009 ) , 0906.3229 .
konchakovski , s. haussler , m.i .
gorenstein , e. l. bratkovskaya , m. bleicher , and h. stcker , phys .
c * 73 * , 034902 ( 2006 ) ; v. p. konchakovski , m. i. gorenstein , e. l. bratkovskaya , h. stcker , phys
c * 74 * , 064911 ( 2006 ) ; v. p. konchakovski , m. i. gorenstein , and e. l. bratkovskaya , phys .
c * 76 * , 031901 ( 2007 ) ; v. p. konchakovski , m. i. gorenstein , and e. l. bratkovskaya , phys .
b * 651 * , 114 ( 2007 ) ; v. p. konchakovski , b. lungwitz , m. i. gorenstein and e. l. bratkovskaya , phys .
c * 78 * , 024906 ( 2008 ) .
|
particle number fluctuations and correlations in nucleus - nucleus collisions at sps and rhic energies have been studied within the hadron - string - dynamics ( hsd ) transport approach .
event - by - event fluctuations of pion - to - kaon , proton - to - pion and kaon - to - proton number ratios are calculated for the samples of most central collision events and compared with the available experimental data .
it has been found that the hsd model can qualitatively reproduce the measured excitation function for the @xmath0 ratio fluctuations in central au+au ( or pb+pb ) collisions from low sps up to top rhic energies .
these predictions impose a challenge for future experiments .
|
the most natural way of defining the entropy of a quantum system is to use the von neumann entropy : @xmath0\ , , \label{entropy : vn}\ ] ] where @xmath1 denotes the density operator which in the schrdinger picture satisfies the von neumann equation : @xmath2 \ , , \label{density op : eom}\ ] ] where @xmath3 is the hamiltonian .
since quantum mechanics and quantum field theory are unitary theories , the von neumann entropy is conserved , albeit in general not zero .
the decoherence program @xcite usually assumes the existence of some environment that is barely observable to some observer .
this allows us to construct a reduced density operator @xmath4 characterising the system s , which is obtained by tracing over the unobservable environmental degrees of freedom e : @xmath5 $ ] .
this is a non - unitary process which consequently generates entropy .
this operation is justified by the observer s inability to see the environmental degrees of freedom ( or better : access the information stored in the es - correlators ) . yet , tracing out some degrees of freedom results in complications .
the simple looking von neumann equation ( [ density op : eom ] ) is replaced by an equation for the reduced density operator .
this `` master equation '' @xcite can be solved for only in very simple situations , which has hampered progress in decoherence studies in the context of interacting quantum field theories . in particular , we are not aware of any known solution of the master equation that would include perturbative corrections and implement the program of renormalisation . also , one has no control of the error in a calculation in the perturbative sense induced by neglecting non - gaussian corrections .
here we propose a decoherence program that can be implemented in a field theoretical setting and that also allows us to incorporate renormalisation procedures .
the idea is very simple , and we present it for a real scalar field @xmath6 . a generalisation to other types of fields , e.g. gauge and fermionic fields , should be quite straightforward .
the density matrix @xmath7 contains all information about the ( possibly mixed ) state a quantum system is in . from the density matrix
one can , in principle , calculate various correlators .
for example , the three gaussian correlators are given by : [ 3 equal time correlators ] @xmath8 = f(\vec{x},t;\vec{y},t')|_{t = t'}\label{3 equal time correlatorsa } \\
\langle \hat{\pi}(\vec{x } ) \hat{\pi}(\vec{y } ) \rangle & = & { \rm tr}[\hat \rho(t ) \hat\pi(\vec x)\hat\pi(\vec y ) ] = \partial_{t } \partial_{t ' } f(\vec{x},t;\vec{y},t')|_{t = t ' } \label{3 equal time correlatorsb } \\ \frac{1}{2 } \langle \ { \hat{\phi}(\vec{x } ) , \hat{\pi}(\vec{y } ) \ }
\rangle & = & \frac12 { \rm tr}[\hat \rho(t)\ { \hat\phi(\vec x),\hat\pi(\vec y)\ } ] = \partial_{t ' } f(\vec{x},t;\vec{y},t')|_{t = t ' } \label{3 equal time correlatorsc}\,,\end{aligned}\ ] ] where @xmath9 denotes the momentum field conjugate to @xmath10 , and where the curly brackets denote the anti - commutator as usual .
note that these three gaussian correlators can all be determined from the statistical propagator @xmath11 in the heisenberg picture as usual in quantum field theory .
the fourth gaussian correlator is trivial as it is restricted by the canonical commutation relations .
there is an infinite number of higher order , non - gaussian correlators that characterise a system , where one can think of e.g. : [ ngcorrelators ] @xmath12 \label{ngcorrelatorsa } \\ \langle \hat{\pi}(\vec{x}_{1 } ) \cdots \hat{\pi}(\vec{x}_{n } ) \rangle & = & { \rm tr}[\hat \rho(t ) \hat\pi(\vec{x}_{1 } ) \cdots \hat\pi(\vec{x}_{n } ) ] \label{ngcorrelatorsb } \,,\end{aligned}\ ] ] where @xmath13 .
moreover , one can also imagine a higher order correlation function consisting of a mixture of both @xmath14 s and @xmath15 s .
finally , one could think of other non - gaussian correlators obtained by further differentiating the statistical propagator : @xmath16 |_{t = t'}\,,\ ] ] where @xmath17 .
for free theories , all the non - gaussian correlators either vanish or can be expressed in terms of the gaussian correlators .
let us turn our attention to interacting field theories and present two simple arguments why , generally speaking , the gaussian correlators dominate over the non - gaussian correlators .
in interacting quantum field theories , the gaussian correlators all stem from tree - level physics , whereas non - gaussian correlators are generated by the interactions . relying on a perturbative treatment of these interactions
, we can safely expect that all of these higher order , non - gaussian correlators above are suppressed as they generically are proportional to a non - zero power of the ( perturbatively small ) interaction coefficient .
a second , more heuristic , argument invokes the ( classical ) central limit theorem , according to which systems with many stochastic mutually independent degrees of freedom tend to become normally distributed , and hence nearly gaussian .
while this theorem is derived for classical systems and for independent stochastic variables , we believe that it will in a certain quantum disguise also apply to weakly interacting quantum systems with many degrees of freedom .
hence , we expect that to a good approximation , many of the relevant properties of quantum systems are encoded in the gaussian part @xmath18 of the density operator @xmath19 and the non - gaussian correlators are thus typically much more difficult to access\sim g_nh^2\sim 10^{-12}$ ] , where @xmath20 denotes the newtonian potential , where @xmath21 is newton s constant and @xmath22 is the hubble rate during inflation , its quantum corrections are expected to be of the order @xmath23 or smaller . ] .
the crucial point is that all information about a general gaussian density matrix and thus about the gaussian part of a state , is stored only in the three equal time correlators ( [ 3 equal time correlators ] ) .
all non - gaussian correlators like ( [ ngcorrelators ] ) contain information about the correlations between the system and environment as well as the environment s correlations induced by higher order interactions . when projected on the field amplitude basis in the schrdinger picture , the gaussian part of a density matrix reads : @xmath24 \equiv \langle \phi| \hat \rho_{\mathrm{g}}(t ) |\phi'\rangle = { \cal n}\exp \ ! \left[\ ! - \ ! \int \ !
\mathrm{d}^3x \mathrm{d}^3y\left ( \phi(\vec x\ , ) a(\vec x,\vec y;t)\phi(\vec y\ , ) + \phi'(\vec x\ , ) b(\vec x,\vec y;t)\phi'(\vec y\,)-2\phi(\vec x\ , ) c(\vec x,\vec y;t)\phi'(\vec y\,)\right ) \right]\ ! , \label{density matrix : gaussian}\ ] ] where @xmath25 and @xmath26 are functions of the equal time field correlators ( for details see sections [ entropy in scalar field theory ] ) .
of course , the von neumann equation ( [ density op : eom ] ) does not hold any more for @xmath27 , and thus the entropy conservation law is violated by @xmath27 . indeed , neglecting the information stored in observationally inaccessible correlators will give rise to an increase of the entropy .
the purpose of the present work is to calculate the entropy of a system taking these considerations into account .
we distinguish two cases . in the simplest case
we assume that our observer can only measure gaussian correlators and his apparatus is insensitive to all non - gaussian correlators .
we will consider this situation in sections [ gaussian entropy from the wigner function ] and [ gaussian entropy from the replica trick ] .
in quantum field theories , the problem seems to be that the evolution equations of the three gaussian correlators ( [ 3 equal time correlators ] ) do not close as soon as perturbative corrections are included .
the reason is that off - shell physics becomes important . a way out of this impasse , and in our opinion _
the _ way out , is to solve for the dynamics of the unequal time statistical correlator , i.e. the statistical propagator , from which one then extracts the three relevant correlators ( [ 3 equal time correlators ] ) . in the second case
, we assume that our observer is sensitive to specific types of non - gaussianities present in our theory , apart from the leading order gaussian correlators of course , either because the observer has developed a sensitive measurement device or the non - gaussianities happen to be large enough .
we will consider this case in section [ non - gaussian entropy : two examples ] .
the sections above contain , for pedagogical reasons , quantum mechanical derivations rather than quantum field theoretical calculations .
we will generalise our results to quantum field theory in section [ entropy in scalar field theory ] .
this paper aims only at developing the necessary machinery to discuss entropy generation in an interacting quantum field theory in an out - of - equilibrium setting , see e.g. @xcite .
we discuss the application of these ideas elsewhere @xcite in great detail , where we consider a scalar field @xmath28 coupled to a second scalar field via the interaction lagrangian density @xmath29 .
the results we develop here are applicable to a much more wider class of field theoretical models , including gauge and fermionic fields .
entropy has of course been widely studied over the years @xcite . in this section
we will derive an entropy formula based on a nave probabilistic interpretation of the wigner function .
this characterises much of the 20th century efforts to connect the wigner function , boltzmann s equations , boltzmann s h - theorem and macroscopic entropy @xcite .
let us begin by considering the hamiltonian of a time dependent harmonic oscillator , which is the one - particle `` free '' hamiltonian : @xmath30 in general , the oscillator s mass and frequency and the source current can all depend on time : @xmath31 , @xmath32 and @xmath33 , respectively .
@xmath34 is the source hamiltonian and @xmath33 the corresponding current .
this hamiltonian is relevant both for many condensed matter systems that can be realised in laboratories and in an early universe setting . for the latter ,
the time dependence of the parameters is introduced for example by the universe s expansion or by phase transitions .
alternatively , this hamiltonian can represent for example the simplest model of a laser , where @xmath35 is the photon amplitude , where , if @xmath36 , @xmath37 is the photon s frequency and where @xmath33 is a charge current that coherently pumps energy into the system , transforming the vacuum state into a coherent state @xcite .
the time dependence of the parameters changes the system s energy , but simultaneously does not change the gaussian nature of a state if such a state is imposed initially , i.e. : the evolution implied by the hamiltonian ( [ hamiltonian:1particle ] ) transforms an initial gaussian density matrix into another gaussian density matrix .
moreover , as we will come to appreciate , one has to add a coupling to the environment or add a self - interaction to generate entropy .
the lagrangian @xmath38 follows as usual from the hamiltonian .
note that the source term can be removed from this lagrangian by a simple coordinate shift : @xmath39 upon which the lagrangian reduces to a quadratic form : @xmath40\,.\ ] ] here , @xmath41 is a @xmath42-independent function of time and the last term is a boundary term which does not contribute to the equation of motion if @xmath43 is a function that obeys : @xmath44+m\omega^{2 } x_{0 } + j = 0 \,.\ ] ] a general gaussian density matrix centered at the new origin remains a gaussian density matrix centered at the origin under the evolution of a quadratic hamiltonian and moreover , which is what the analysis above shows , linear source terms in the hamiltonian will not alter the gaussian nature of the state .
we can thus consider gaussian density matrices centered at the origin , whose time evolution is governed by an hamiltonian of the form ( [ hamiltonian:1particle ] ) with @xmath45 .
when written in position space representation , the single particle density operator of a quantum mechanical gaussian state centered at the origin is of the form : [ density operator : particle ] @xmath46 where : @xmath47 \,,\ ] ] and where @xmath48 , @xmath49 and @xmath50 are determined from the von neumann equation ( [ density op : eom ] ) .
the subscript g denotes `` gaussian '' .
moreover , from @xmath51 it follows that @xmath52 and @xmath53 . when @xmath54 one recovers a pure state with vanishing entropy .
however when @xmath55 the density matrix is mixed and entangled and it can not be written in the simple diagonal form , @xmath56 , where @xmath57 . when @xmath58 the density matrix tends to get diagonal in the @xmath59 direction , however when @xmath60 the diagonalisation occurs in the @xmath61 direction in which case the entropy is not defined . the heisenberg uncertainty relation is in trouble in this case as we will see in equation ( [ uncertainly relation ] ) .
hence it is natural to assume that @xmath58 .
we keep the discussion phenomenological and do not discuss the physical origin of @xmath55 .
the normalisation @xmath26 is obtained from requiring @xmath62=1 $ ] : @xmath63 = \int_{-\infty}^{\infty } \mathrm{d } \tilde{x } \langle \tilde x| \hat \rho_{\mathrm{g } } | \tilde x \rangle = \int_{-\infty}^{\infty } \mathrm{d}x \rho_{\mathrm{g}}(x , x;t ) = { \cal n } \sqrt{\frac{\pi}{2(a_{\mathrm{r}}-c)}}= 1 \,,\ ] ] from which we conclude : @xmath64 provided that @xmath65 where @xmath66 $ ] . the equations that the functions @xmath67 , @xmath68 and @xmath69 of the density matrix ( [ density operator : particle2 ] ) obey can easily be obtained from the von neumann equation ( [ density op : eom ] ) , which in this gaussian case in the amplitude basis reads : @xmath70 if we insert equation ( [ density operator : particle2 ] ) in the equation above we find : [ density matrix : eom:2 ] @xmath71 where we defined @xmath72 $ ] .
note that @xmath73 is consistent with equation ( [ density matrix : eom:2c ] ) as it should ) one can show that equations ( [ density matrix : eom:2 ] ) are consistent with ( and equivalent to ) the hamilton equations for the correlators ( [ hamilton equations : correlators ] ) . ] .
the wigner function is defined as a wigner transform of the density matrix @xmath74 : @xmath75 where we defined the average and relative coordinates , @xmath76 and @xmath77 , respectively .
hence , a wigner transform can be thought of as an ordinary fourier transform with respect to the relative coordinate @xmath78 of the density matrix .
the inverse wigner transform determines @xmath74 in terms of @xmath79 : @xmath80 where @xmath81 is the standard phase space measure .
we obtain the gaussian wigner function by performing the integral ( [ wigner function : def ] ) : @xmath82 \,,\ ] ] where : [ wigner function : gauss2 ] @xmath83 and @xmath84 denotes a `` classical '' momentum , which makes the wigner function non - diagonal .
note that the wigner function ( [ wigner function : gauss ] ) is normalised to unity , in the sense that : @xmath85 recalling definition ( [ wigner function : def ] ) , equation ( [ density matrix : eom ] ) transforms to : @xmath86 where we made use of the following relations : [ intermediatestep1 ] @xmath87 upon noting that @xmath88 and @xmath89 , we see that equation ( [ wigner function : eom ] ) is nothing but the liouville equation for the boltzmann s distribution function @xmath90 , @xmath91 a probabilistic interpretation of the wigner function thus seems plausible by the following identification : @xmath92 of course , the equations for @xmath93 and @xmath90 are identical only in a free harmonic oscillator theory and , when interactions are included , differences arise between the von neumann equation for the density operator @xmath94 ( or the corresponding equation for @xmath79 ) and the boltzmann equation for @xmath90 . this indeed makes the identification ( [ w = f ] ) less rigorous but it remains a useful picture .
the correspondence ( [ w = f ] ) has been widely used in the literature to develop a formalism relevant for example for the physics of heavy ion collisions @xcite or for baryogenesis @xcite . in the distant 1969 cahill and glauber @xcite
have proposed to approximate the density operator in the coherent state basis @xmath95 by its diagonal form : @xmath96 , which they interpret as a quasi - probability distribution . since coherent states harbour many properties of classical particles
, one can argue that this is reasonable . recall that coherent states are defined by @xmath97 , where @xmath98 $ ] denotes the annihilation operator for the oscillator ( [ hamiltonian:1particle ] ) .
the density matrix @xmath99 represents an early example of a quasi - probability distribution for a quantum system .
needless to say , by this line of reasoning one can get only approximate answers for the expectation value @xmath100 $ ] of an operator @xmath101 and likewise for the entropy ( which is just the expectation value of @xmath102 ) .
another common way of approximating expectation values of operators is to give a probabilistic interpretation to the wigner function , along the same line of reasoning as argued above in equation ( [ w = f ] ) for a free theory . in this approach @xcite the entropy is approximated by : @xmath103 \ , , \label{entropy : wigner}\ ] ] where the trace @xmath104 $ ] should be interpreted as an integration over the phase space volume as in equation ( [ trw=1 ] ) .
we now insert ( [ wigner function : gauss ] ) into ( [ entropy : wigner ] ) to find : @xmath105 where we defined : @xmath106 we evaluated the phase space gaussian integrals by shifting the momentum integration to @xmath107 ( the jacobian of this shift equals unity ) . at this point
, it is useful to evaluate a few quantum mechanical expectation values .
we can extract the following correlators from our gaussian state : [ correlators : all ] @xmath108 = \int_{-\infty}^{\infty } \mathrm{d}\tilde{x }
\langle \tilde{x}| \hat\rho_{\mathrm{g}}\hat x^2 | \tilde{x}\rangle = \frac{1}{4(a_{\mathrm{r } } -c ) } \label{correlators : alla}\\ \big\langle \frac12\{\hat x,\hat p\}\big\rangle & = & -\hbar\frac{a_{\mathrm{i } } } { 2(a_{\mathrm{r } } -c ) } \label{correlators : allb}\\ \langle \hat p^2\rangle & = & \hbar^2\frac{|a|^2-c^2}{a_{\mathrm{r } } -c } \ , .
\label{correlators : allc}\end{aligned}\ ] ] here , we made use of @xmath109 .
moreover , one can easily verify that @xmath110\rangle = \langle \hat x\hat p-\hat p\hat x\rangle = \imath\hbar $ ] as it should .
this enables us to find the equivalent inverse relations : [ ai : ar : c ] @xmath111 and to express @xmath112 in equation ( [ delta function:2 ] ) in terms of the correlators ( [ correlators : all ] ) : @xmath113 \ , .
\label{delta function:3}\ ] ] for the moment this is just a definition , but as we will come to discuss , the physical meaning of @xmath114 is the phase space area in units of @xmath115 occupied by a ( gaussian ) state centered at the origin . for a pure state @xmath116 , while for a mixed state @xmath117 .
hence , we can define the uncertainty relation for a general gaussian state : @xmath118 if @xmath119 is negative and @xmath120 , we can in principle have @xmath121 , which can be seen from equation ( [ delta function:2 ] )
. it would be interesting to investigate whether any physical experiment could be proposed where a violation of heisenberg s uncertainty relation occurs , even if it were for only a very brief moment in time ( such that it would still hold on average ) .
furthermore , it is natural to define the statistical particle number density on our phase space , in the quantum field theoretical sense , as : @xmath122 in terms of which the wigner entropy ( [ wigner function : entropy : final ] ) can be rewritten as : @xmath123 these results in essence agree with the entropy per mode of scalar cosmological perturbations obtained in @xcite .
shortly we shall see that the wigner entropy ( [ wigner function : entropy:2 ] ) is a good approximation for the von neumann entropy if @xmath124 , or equivalently when @xmath125 .
let us try to put these results briefly in historical context .
the fundamental question that is left unanswered in @xcite is what the dynamical mechanism for the growth of @xmath126 is . instead
, the authors argue that an observer that measures cosmological perturbations can be approximated by a coherent state ( which is a good classical basis ) , thus explaining the growth in @xmath126 .
two mechanisms were considered in @xcite : a projection on the particle number basis and on the coherent state basis . in both cases , information of the phase of the squeezed gaussian state
is lost , which thus generates entropy . in the latter approach formula ( [ delta function:3 ] )
is reproduced , while in the former , the entropy acquires the form of the statistical entropy of bose particles @xmath127 with @xmath128 the particle number associated with the squeezed state , and @xmath78 the squeezing parameter of the state . while these early articles , and many subsequent ones @xcite , attempted to explain the origin of entropy of cosmological perturbations by a ( non - unitary ) process of coarse graining ( associated with particular properties of the observer s measuring apparatus ) , a genuine dynamical mechanism of entropy generation was lacking .
this has changed recently @xcite , now dynamical mechanisms for entropy generation have been proposed . rather than discussing various mechanisms in detail ,
let us just point out that the notion of how entropy should be defined and generated has evolved : recently , it has been advocated that the relevant entropy to consider is the gaussian entropy , although the approaches differ in the actual origin of the entropy increase . for example ,
prokopec and rigopoulos suggested in @xcite that tracing over the unobserved isocurvature perturbations leads to entropy generation in the adiabatic mode .
similarly , tracing over the unobserved gravitational waves should lead to entropy generation in the adiabatic mode @xcite .
kiefer , lohmar , polarski and starobinsky @xcite as well as campo and parentani @xcite , among others , suggested that , if super - hubble modes take the role of the system , the origin of the entropy of cosmological perturbations should be associated with the stochastic noise generated by the sub - hubble ultraviolet modes representing the environment . both giraud and serrau @xcite and we @xcite advocate that neglecting observationally inaccessible non - gaussianities is a prominent mechanism to generate entropy . .
[ fig : gaussianphasespace2 ] ] _ _ .1 cm .
[ fig : gaussianphasespace2 ] ] _ _ .
[ fig : gaussianphasespace2 ] ] _ _ so far , we have discussed how the gaussian entropy in wigner space is defined and qualitatively discussed various mechanisms to generate entropy dynamically .
let us now , to finish this section , discuss how one can sensibly think of a decohered state in wigner space and discuss the physical meaning of @xmath112 .
despite the fact that the wigner entropy ( [ wigner function : entropy : final ] ) only approximates the von neumann entropy , it offers a useful , intuitive way of depicting decohered states irrespective of the precise underlying mechanism by which such a state has decohered in the first place .
let us , for this reason , discuss this phenomenologically .
the wigner function ( [ wigner function : gauss ] ) is a 2-dimensional gaussian function whose width in the @xmath129-direction need a priori not necessarily be the same as in the @xmath130-direction .
hence , in general it can be squeezed in some direction .
this is illustrated in figure [ fig : gaussianphasespace3 ] .
the @xmath131 or @xmath132 cross - section of the phase space distribution has the shape of an ellipse , which we also visualised in figure [ fig : gaussianphasespace3 ] .
if the cross - section is a circle centered at the origin , the state is either pure or mixed and if , moreover , its area is unity @xmath133 it is a pure vacuum state .
if the cross - section is again a circle but displaced from the origin , we are dealing with a generalised coherent - squeezed state ( whose shape is not described by equation ( [ wigner function : entropy : final ] ) for obvious reasons ) . if the cross - section is an ellipse , it is a squeezed state as mentioned before .
the ellipse is parametrised by : @xmath134 where we have made use of equation ( [ wigner function : gauss ] ) and where @xmath135 is some constant that determines the height at which we slice the wigner function .
the area of this ellipse is now given by is @xmath136 . ] : @xmath137 where we have used equations ( [ wigner function : gauss2 ] ) and ( [ delta function:2 ] ) .
clearly , the area of the ellipse in wigner space is determined straightforwardly from the phase space area @xmath112 .
note that if @xmath138 we have @xmath139 .
figures [ fig : gaussianphasespace1 ] and [ fig : gaussianphasespace2 ] visualise the decoherence process of a squeezed state .
initially , in figure [ fig : gaussianphasespace1 ] , the state is squeezed and has unity phase space area @xmath140 , such that its entropy vanishes . now we switch on our favourite decoherence mechanism : the state interacts with some environment and environmental degrees of freedom are traced over or non - gaussian correlators generated by the interaction with the environment are neglected .
consequently , the area in phase space increases @xmath117 , such that @xmath141 .
hence , the ellipse in wigner space grows which we depict in figure [ fig : gaussianphasespace2 ] .
an important feature of any decoherence process reveals itself : for a highly squeezed state , given the knowledge of @xmath130 after some measurement , the value @xmath129 can take in a subsequent measurement is constraint in a very narrow interval .
interestingly , after the state has decohered , and given the same measurement of @xmath130 , the range of values @xmath129 can take has increased .
indeed one can say that knowledge about @xmath129 has been lost compared to the squeezed state before decoherence , hence entropy has been generated .
alternatively , we can say that given some measurement of @xmath130 , the value @xmath129 can take is , after decoherence , drawn from a classical stochastic distribution , which indeed coincides with the familiar idea that decohered quantum systems should behave as uncorrelated stochastic systems .
indeed , @xmath142 roughly counts the number of patches in phase space that behave independently and are uncorrelated . with this intuitive notion of decohered state in mind , one can understand that , for highly squeezed states , the position space basis ( which generalises to the field amplitude basis in the quantum field theoretical case ) is the pointer basis .
let us start out with a highly squeezed state , i.e. : take for example a pure state during inflation that rapidly squeezes during the universe s expansion .
the hamiltonian of the state is then dominated by the potential term , and the kinetic term only contributes little . in thermal equilibrium , the system will minimise its free energy @xmath143 , where @xmath144 is the temperature of the heat bath .
now , if we switch on some decoherence mechanism due to interaction with the environment , the entropy will increase mainly due to momentum increase , i.e. : @xmath145 will increase .
indeed , increasing @xmath145 will hardly affect the hamiltonian , whereas it will significantly affect the entropy .
hence , the variable @xmath146 is robust during the process of decoherence , in the sense that @xmath147 will hardly change such that it qualifies as a proper pointer basis .
note however , that @xmath146 is only a pointer basis in the statistical sense , such that there is a well defined probability distribution function from which a measurement can be drawn .
the entropy of a general gaussian state has been derived by sohma , holevo and hirota @xcite by making use of glauber s p representation for the density operator .
here we present an alternative derivation for the von neumann entropy ( [ entropy : vn ] ) of a general gaussian state by making use of the replica trick ( see e.g. : @xcite ) .
of course , there are other analogous methods one can use .
notice first that : @xmath148 here , @xmath149 is a positive integer which is analytically extended to zero .
hence , in order to calculate the entropy ( [ entropy : vn ] ) we need to evaluate @xmath150 $ ] . using equation ( [ density operator : particle ] )
, we thus have : @xmath151 & = & { \cal n}^{1+\epsilon } \int_{-\infty}^{\infty } \mathrm{d}x_0 \cdots \int_{-\infty}^{\infty } \mathrm{d}x_{\epsilon}\exp \left[-2a_{\mathrm{r } } \left(x_0 ^ 2+\cdots + x_\epsilon^2 \right ) + 2c \left(x_0x_1 + x_1x_2 + \cdots + x_\epsilon x_0 \right ) \right ] \nonumber\\ & = & { \cal n}^{1+\epsilon}\int_{-\infty}^{\infty } \mathrm{d}x_0 \cdots \int_{-\infty}^{\infty } \mathrm{d}x_{\epsilon}\exp \left[-2\beta \left\ { ( x_0 - \alpha x_1)^2+(x_1 - \alpha x_2)^2 + \cdots + ( x_\epsilon - \alpha x_0)^2 \right\ } \right ] \nonumber\\ & = & |1-\alpha^{1+\epsilon}|^{-1 } \left({\cal n}\int_{-\infty}^{\infty } \mathrm{d}y \exp\left [ -2\beta y^2 \right ] \right)^{1+\epsilon } \nonumber\\ & = & |1-\alpha^{1+\epsilon}|^{-1 } \left(\frac{a_{\mathrm{r}}-c}{\beta}\right)^{\frac{1+\epsilon}{2}}\ , .
\label{replica trick:2}\end{aligned}\ ] ] here , @xmath152 is the jacobian of the transformation from @xmath153 to @xmath154 , where @xmath155 , and @xmath156 .
a comparison of the first and the second line in ( [ replica trick:2 ] ) tells us that : [ replica trick:3 ] @xmath157 in the limit @xmath158 one must recover a pure diagonal density operator and , consequently , the shift @xmath159 should vanish , singling out : [ replica trick:4 ] @xmath160 in equation ( [ replica trick:3 ] ) as the physical choice .
based on @xmath161 , we can rewrite equation ( [ replica trick:2 ] ) in a simpler form : @xmath162 = \frac{(1-\alpha)^{1+\epsilon}}{|1-\alpha^{1+\epsilon}| } \ , .
\label{replica trick:5}\ ] ] recall that normalisability of the density operator requires @xmath163 , implying that @xmath164 , such that the absolute value in ( [ replica trick:5 ] ) can be dropped .
this means that @xmath150 $ ] is analytic in @xmath159 ( for positive integers @xmath149 ) : @xmath162 = \frac{(1-\alpha)^{1+\epsilon}}{1-\alpha^{1+\epsilon } } \ , , \label{replica trick:6}\ ] ] such that one can analytically extend to complex @xmath149 .
according to the replica method , the entropy is then obtained by taking the limit @xmath165 . to perform this procedure
we expand ( [ replica trick:6 ] ) to linear order in @xmath149 and we find the following expression for the von neumann entropy ( [ entropy : vn ] ) : @xmath166 using the expression for @xmath159 in equation ( [ replica trick:4a ] ) and the definition of the phase space area in equations ( [ delta function:2 ] ) and ( [ delta function:3 ] ) , we find : @xmath167 such that we can express the entropy ( [ entropy : rho log rho : final ] ) solely in terms of @xmath126 : @xmath168 since @xmath169 , we have @xmath170 .
the von neumann entropy ( [ entropy : final ] ) vanishes when @xmath171 ( or @xmath54 ) , defining a pure state , while a strictly positive entropy @xmath172 implies a mixed state @xmath173 .
clearly , the gaussian von neumann entropy is completely determined by the three gaussian correlators characterising the state ( [ delta function:3 ] ) .
relations ( [ entropy : final ] ) and ( [ entropy : rho log rho : final ] ) suggest the following definition of the particle number @xmath174 : @xmath175 where @xmath176 , in terms of which the entropy ( [ entropy : rho log rho : final ] ) , ( [ entropy : final ] ) becomes : @xmath177 which is the well known result from statistical physics for the entropy of @xmath142 bose particles per ( quantum ) state @xcite .
note that ( [ entropy : n ] ) is a convex function of @xmath142 , such that @xmath178 , which is another desirable property for the entropy .
the particle number @xmath142 defined in ( [ n ] ) should be interpreted as the number of independent ( uncorrelated ) regions in the phase space of a single ( one particle ) quantum state .
we will show next that @xmath126 is a special function , as it is conserved by the evolution equations resulting from the hamiltonian ( [ hamiltonian:1particle ] ) : [ hamilton equations ] @xmath179 where we set @xmath180 .
these equations imply for the correlators : [ hamilton equations : correlators ] @xmath181\rangle & = & 0 \label{hamilton equations : correlatorsc}\\ \frac{\mathrm{d}}{\mathrm{d}t}\langle\hat p^2\rangle & = & - 2m\omega^2\big\langle\frac12\{\hat x,\hat p\}\big\rangle \label{hamilton equations : correlatorsd}\,.\end{aligned}\ ] ] recall that these equations are , in the light of equation ( [ ai : ar : c ] ) , equivalent to ( [ density matrix : eom:2 ] ) .
the third equation is automatically satisfied by the commutation relations , @xmath182 = \imath \hbar$ ] .
one can combine the other three equations in ( [ hamilton equations : correlators ] ) to show that : @xmath183 = 0 \label{delta : conservation}\ ] ] this implies that @xmath126 ( and in fact any function of @xmath126 ) is conserved by the hamiltonian evolution ( [ hamilton equations][hamilton equations : correlators ] ) .
of course , when interactions are included , the conservation law ( [ delta : conservation ] ) becomes approximate . in quantum field theory interactions
are typically cubic or quartic in the fields and will thus induce non - gaussianities in the density matrix .
in particular , for a quantum mechanical @xmath184 model , this has been investigated by calzetta and hu @xcite .
they derive an h - theorem for @xmath126 in the case of a quantum mechanical @xmath185 model . .
right : the difference between the wigner and the von neumann entropy .
the wigner entropy approaches the exact von neumann entropy rapidly as @xmath126 grows .
[ fig : entropycomp ] ] .
right : the difference between the wigner and the von neumann entropy . the wigner entropy approaches the exact von neumann entropy rapidly as @xmath126 grows .
[ fig : entropycomp ] ] _ _ finally , it is interesting to compare the von neumann entropy ( [ entropy : final ] ) and the wigner entropy ( [ wigner function : entropy : final ] ) , as shown in figure [ fig : entropycomp ] .
this reveals that the wigner entropy represents a large phase space ( semiclassical ) limit of the von neumann entropy , which should not come as a surprise .
indeed , the wigner entropy can be used to accurately represent the entropy in systems which develop large correlators ( occupation numbers ) , such as cosmological perturbations @xcite .
the wigner entropy fails however to take account of quantum correlations present in the density matrix and wigner function , which are properly taken into account in the exact expression for the entropy of a general gaussian state ( [ entropy : final ] ) .
an important question is how to generalise the von neumann entropy of a gaussian state to include non - gaussian corrections .
non - gaussianities can be important either because they are made large on purpose by specially preparing the state , or simply because they are accessible in measurements due to the fact that the observer s measurement device is very accurate .
while we are far from building a general theory of how non - gaussian correlations affect the notion of phase space volume , statistical particle number and thus entropy , we shall present two examples in this section : we consider a system in a state which can be represented by an admixture of a gaussian density matrix and some small non - gaussian contributions . secondly , we consider small quartic corrections to the density matrix , contributing to the kurtosis of the ground state .
qualitatively , we expect that the correction to the gaussian entropy due to non - gaussianities present in a theory is small , except for systems whose entropy is ( almost ) zero .
this can most easily be appreciated in the wigner approach to entropy .
consider for example a system with a large entropy , for example a squeezed state whose phase space area in wigner space has increased significantly due to developing a @xmath186 .
non - gaussianities can deform the area in phase space of the state ( by squeezing , pushing or stretching the wigner function ) roughly by a factor of order unity .
consequently , the effect on the entropy will be relatively small .
this can be understood by realising that a lot of information is contained in unequal time correlators .
of course , such an argument ceases to be true for a pure state , whose quantum properties are very pronounced .
let us consider a density matrix of the form : @xmath187 such a state typically appears when one uses a laser to excite an harmonic oscillator in its ground state .
the system can then be described by an admixture of the ground state and a small contribution to the first exited state ) ( with @xmath188 ) is : @xmath189^{1/4 } \ , x\ , \exp\big(-\frac{m\omega x^2}{2\hbar}\big ) \ , , \nonumber\ ] ] with energy @xmath190 .
the problem at hand reduces to this state upon identifying @xmath191 , @xmath192 , @xmath158 , @xmath193 and @xmath194 . ] .
note that the process of pumping energy in a simple harmonic oscillator in its ground state could lead to creating a state that is not pure ( @xmath195 ) .
the measuring apparatus is assumed to be sensitive to the admixture of the two states .
we shall assume that non - gaussianity is small , in the sense that all @xmath196 parameters ( @xmath197 ) are small , such that we content ourselves with performing the analysis up to linear order in @xmath196 .
notice first that to this order the terms in parentheses in equation ( [ rho2 ] ) can be approximated by an exponential : @xmath198 implying that the first two terms @xmath199 induce an entropy conserving shift in @xmath146 and @xmath200 ( as in coherent states ) .
of course , this ceases to be true at quadratic and higher orders in @xmath201 . on the other hand ,
the @xmath202 and @xmath203 terms do change the entropy even at linear order . at higher order , @xmath202 and @xmath203 induce kurtosis whose effect on the entropy we will discuss shortly . before we turn our attention to calculating the entropy ,
let us first show that a density matrix ( [ rho2 ] ) with @xmath204 indeed generates non - trivial higher order correlators .
normalising the trace to unity yields : @xmath205 an interesting higher order correlator to consider is for example the connected part of the four - point correlator : @xmath206 if the connected part of the four - point correlator is non - zero , the state is said to have kurtosis . using the gaussian density matrix in equation ( [ density operator : particle ] )
one can easily show that the connected part of the four - point correlator vanishes as it should : for free theories higher order correlators either vanish or can be expressed in terms of gaussian correlators .
using however the non - gaussian density matrix in equation ( [ rho2 ] ) with @xmath207 and the normalisation constant ( [ normdensity2 ] ) we find : @xmath208 clearly , genuine non - gaussian correlators are generated and this state has kurtosis at quadratic order .
let us now calculate the entropy by making use of equation ( [ exponentialassumption ] ) .
as said , it is precisely @xmath209 and @xmath210 that affect the entropy . to see that notice further that their effect can be captured by a shift in @xmath211 and @xmath119 as follows : [ a+c : rescaling ] @xmath212 such that the corresponding von neumann entropy simply reads ( _ cf .
_ equation ( [ entropy : n ] ) ) : @xmath213 where @xmath214 and where as usual @xmath215 .
now @xmath216 is the correction to the gaussian entropy we are about to calculate . comparing with equations ( [ replica trick:3 ] ) and ( [ n ] ) we have : [ nandalphabar ] @xmath217 from which we immediately find , to linear order in @xmath218 and @xmath210 : [ deltananddeltaalpha ] @xmath219 where we used @xmath220 .
finally , @xmath221\}\delta n$ ] implies that : [ s2:final ] @xmath222 where @xmath223 is an entropy shift given by : @xmath224 where @xmath225 is the von neumann entropy of a gaussian state and @xmath142 is the statistical particle number associated with the gaussian part of the state as before .
several comments are in order .
firstly , the result ( [ s2:final ] ) implies that to linear order @xmath226 $ ] does not change the entropy .
in fact , @xmath226 $ ] induces ( de)squeezing of the state , which can be appreciated from equation ( [ wigner function : gauss2c ] ) . secondly
, @xmath227 is positive ( negative ) whenever @xmath218 and @xmath203 are positive ( negative ) , irrespective of @xmath228 .
thirdly , we have rescaled @xmath218 and @xmath203 by @xmath229 to get a dimensionless quantity .
this rescaling is natural , since @xmath230 measures the width of the state .
finally , for large @xmath142 the formula ( [ s2:final ] ) gives a meaningful answer , since : @xmath231 is finite . on the other hand , when @xmath232 , we encounter a logarithmic divergence : @xmath233 indicating a mild ( logarithmic ) breakdown of the linear expansion .
notice however that also in that limit the formula ( [ s2 ] ) remains applicable . in the former example non
- gaussianities are generated by adding a quadratic polynomial to the prefactor .
one could also introduce non - gaussianities by adding higher order , i.e. : cubic or quartic , powers in the exponential .
cubic corrections generate skewness and since they contribute to the entropy at quadratic and higher orders in the skewness parameter , we shall now focus on the quartic corrections to the density matrix .
quartic corrections to a density matrix affect the kurtosis of a state and we parametrise this as follows : @xmath234 where @xmath235 is the gaussian density matrix ( [ density operator : particle ] ) as before , with the normalisation @xmath236 given in equation ( [ norm of rho:2 ] ) , and where finally @xmath237 . as in the previous example
, we shall consider only linear corrections in @xmath238 ( @xmath197 ) to the entropy . to this order the non - gaussian part of ( [ rho4 ] )
can also be written as : @xmath239 we can find the normalisation constant after some simple algebra by making use of ( [ rhong : lin ] ) : @xmath240 ^ 2 } \bigg]^{-1 } \ , .
\label{caln4}\ ] ] ) and ( [ rhong : lin ] ) , with @xmath241 only .
we used @xmath242 ( left ) and @xmath243 ( right ) .
the peak structure of our state in phase space , that was so simple for gaussian density matrices , is much more complicated .
some regions in phase space have a negative value of the wigner function , indicating a breakdown of the use of a wigner function as a convenient measure of probability .
[ fig : ngwigner ] ] ) and ( [ rhong : lin ] ) , with @xmath241 only .
we used @xmath242 ( left ) and @xmath243 ( right ) .
the peak structure of our state in phase space , that was so simple for gaussian density matrices , is much more complicated .
some regions in phase space have a negative value of the wigner function , indicating a breakdown of the use of a wigner function as a convenient measure of probability .
[ fig : ngwigner ] ] _ _ to gain some intuitive understanding of what our density matrix looks like , let us consider figure [ fig : ngwigner ] . here
, we show the wigner transform of equation ( [ rho4 ] ) using ( [ rhong : lin ] ) , with @xmath244 .
clearly , the quartic corrections change the peak structure of our state in phase space .
moreover , some regions in phase space now have a negative wigner function .
this nicely illustrates the limitations of using the wigner function as a probability density on our phase space as soon as non - gaussianities ( due to interactions ) are included , i.e. : equation ( [ w = f ] ) holds only approximately in this case . in order to calculate the entropy
, we need @xmath245 $ ] as before ( _ cf . _ equation ( [ replica trick:2 ] ) ) . to linear order in @xmath246
we have : @xmath247 = \left(\frac{{\cal n}_4}{{\cal n}_{\rm g}}\right)^{1+\epsilon } \big ( { \rm tr}[\hat \rho_{\rm g}^{1+\epsilon } ] + ( 1+\epsilon){\rm tr}[\hat \rho_{\rm g}^{1+\epsilon}\delta\hat\rho_{\rm ng } ] \big ) \ , .
\label{rho4:lin}\ ] ] the first term in the parentheses is just the gaussian result ( [ replica trick:2 ] ) while the latter term can be evaluated by making use of the formulae in appendix [ appendix a useful integrals ] . expanding to linear order in @xmath149 , the equation above yields : @xmath248 & = & 1 + \epsilon\big[\ln(1-\alpha)+\frac{\alpha}{1-\alpha}\ln(\alpha)\big ] + \epsilon \bigg [ \frac34\frac{\eta_0+\eta_0^*}{[2(a_{\mathrm{r}}-c)]^2 }
\big(\frac{4\alpha}{1-\alpha^2}\ln(\alpha)-1\big ) \nonumber\\ & & \hskip 1 cm + \,\frac34\frac{\eta_1+\eta_1^*}{[2(a_{\mathrm{r}}-c)]^2 } \big(\frac{1+\alpha}{1-\alpha}\ln(\alpha)-1\big ) + \frac34\frac{\eta_2}{[2(a_{\mathrm{r}}-c)]^2 } \big(\frac43\frac{1+\alpha+\alpha^2}{1-\alpha^2}\ln(\alpha)-1\big ) \bigg ] \,.\qquad \label{rho4:lin : b}\end{aligned}\ ] ] as expected , we see that the entropy naturally splits into a gaussian and a non - gaussian contribution : [ deltas4 ] @xmath249 where @xmath250 is the gaussian entropy as before , and where : @xmath251 ^ 2 } \left(\frac{3n(1+n)}{1 + 2n}\ln\left(\frac{1+n}{n}\right)+\frac34\right ) + \frac34\frac{\eta_1+\eta_1^*}{[2(a_{\mathrm{r}}\!-\!c)]^2 } \left((1 + 2n)\ln\left(\frac{1+n}{n}\right)+1\right ) \nonumber\\ & & + \,\frac{\eta_2}{[2(a_{\mathrm{r}}\!-\!c)]^2 } \left ( \frac{1 + 3n+3n^2}{1 + 2n}\ln\left(\frac{1+n}{n}\right)+\frac34\right ) \ , . \label{deltas4b}\end{aligned}\ ] ] this is the main result of this section and intuitive in the following sense : just as in the first non - gaussian example above , we see that positive kurtosis parameters @xmath252 , tend to make the effective state s width @xmath253 larger , which increases the area in phase space the state occupies , which in turn increases the entropy .
if however @xmath254 , the phase space area shrinks , which in turn decreases the entropy . the statements above hold for any statistical particle number @xmath255 , with the exception of @xmath232 , where , just as in the case studied above , a weak logarithmic divergence occurs when @xmath256 or @xmath257 .
notice finally that @xmath258 $ ] and @xmath259 $ ] again do not participate in entropy generation , but rather contribute to the squeezing of the state .
kurtosis and skewness in quantum mechanics , studied in this section , occur also in interacting quantum field theories which we discuss next .
the quantum mechanical expressions for the gaussian and non - gaussian entropies that we have developed for pedagogical reasons in sections [ gaussian entropy from the wigner function ] , [ gaussian entropy from the replica trick ] and [ non - gaussian entropy : two examples ] for a single particle density matrix can be generalised to field theory .
we firstly need to consider correlators in quantum field theory however .
let us now proceed analogous to equation ( [ density operator : particle ] ) and write the density matrix operator for our system in the field amplitude basis in schrdinger s picture ( see e.g. @xcite ) : @xmath260\langle \phi^\prime| \,,\ ] ] where : @xmath261 = { \cal n } \exp\left(-\phi^t\!\cdot a\cdot\phi - { \phi^\prime}^t\!\cdot b\!\cdot\!\phi^\prime + 2\phi^t\!\cdot c\!\cdot\!\phi^\prime \right ) \,,\ ] ] where @xmath262 and @xmath263 .
a few words on the notation first .
we shall consider @xmath264 as a vector whose components are labelled by @xmath265 .
moreover , @xmath266 can be viewed as a matrix , such that @xmath267 is a vector again , where a @xmath268 denotes matrix multiplication which , for the case at hand , is nothing but an integral over @xmath269 dimensional space .
hence , quantities like @xmath270 are scalars and involve two integrals over space .
note that at this point we do not assume that @xmath271 is homogeneous , i.e. : @xmath272 , but rather keep any possible off - diagonal terms for generality .
our density matrix ( [ density operator : field2 ] ) is hermitian , such that we have @xmath273 and @xmath274 , where we also used @xmath275 and @xmath276 .
the normalisation @xmath26 of @xmath277 $ ] can be determined from the standard requirement @xmath278=1 $ ] : @xmath279\right)^{1/2 } \
, , \label{normrhofunc}\ ] ] where we note that the hermitian part @xmath280 of @xmath271 is real and symmetric such that @xmath281 . just as in the quantum mechanical case , we need to calculate the three non - trivial gaussian correlators which completely characterise the properties of our gaussian state . in order to calculate these correlators , we need to have an expression for the statistical propagator as in equation ( [ 3 equal time correlators ] ) which , given some initial density matrix @xmath282 , is in the heisenberg picture defined by : @xmath283 \label{statisticalpropagator } \,.\ ] ] let us begin by calculating : @xmath284 \label{f:1 } \,,\ ] ] where we have made use of the heisenberg evolution equation for operators .
it is convenient to add a source current @xmath285 to the density matrix ( [ density operator : field ] ) , such that @xmath286 becomes : @xmath287
= { \cal n } \exp\left(-\phi^t\!\cdot a\cdot\phi - { \phi^\prime}^t\!\cdot a^{\dag}\!\cdot\!\phi^\prime + 2\phi^t\!\cdot c\!\cdot\!\phi^\prime + j^t\!\cdot \phi + { j'}^t\!\cdot \phi^\prime \right ) \ , , \label{density operator : field : j}\ ] ] in terms of which equation ( [ f:1 ] ) can be rewritten as : [ correlatorsqft ] @xmath288\big|_{j = j^\prime=0 } \nonumber\\ & = & \frac{\delta}{\delta j(\vec x\,)}\frac{\delta}{\delta j(\vec y\,)}\int { \cal d}\tilde\phi \ , { \cal n } \left .
\exp\left(-2\tilde\phi^t\!\cdot(a_{\mathrm{r}}-c)\cdot\tilde \phi\right)\exp\left(\frac18(j+j^\prime)^t\!\cdot(a_{\mathrm{r}}-c)^{-1 } \cdot(j+j^\prime)\right ) \right|_{j = j^\prime=0 } \nonumber\\ & = & \frac14(a_{\mathrm{r}}-c)^{-1}(\vec x,\vec y;t)\ , .
\label{correlatorsqfta}\end{aligned}\ ] ] we also need the other correlators : @xmath289 = -\frac{\hbar}{2}(a_{\mathrm{r}}-c)^{-1}\cdot a_{\mathrm{i}}(\vec x,\vec y;t ) \label{correlatorsqftb } \\
\frac{1}{2 } \langle \ { \hat{\pi}(\vec{x } ) , \hat{\pi}(\vec{y } ) \ } \rangle & = & \partial_t\partial_{t^\prime}f(\vec x , t;\vec y , t^\prime)|_{t = t^\prime } = \frac12{\rm tr}[\hat\rho_{\mathrm{g } } ( t ) \{\hat\pi(\vec x\,),\hat\pi(\vec y\,)\ } ] \label{correlatorsqftc}\\ & = & \hbar^2 \ !
\left [ \frac{1}{2 } a^{\dag } \ !
\cdot\ ! ( a_{\mathrm{r } } \ !
-c)^{-1 } \ !
\cdot \!a \!+\
! a \cdot\ ! ( a_{\mathrm{r}}\!-c)^{-1 } \
! \cdot a^{\dag}\ ! -c \!\cdot\ !
( a_{\mathrm{r}}\!-c)^{-1}\ !
\cdot c \!\right](\vec x,\vec y;t ) \nonumber\,.\end{aligned}\ ] ] we have moreover made use of @xmath290 $ ] . as a check one can verify that : @xmath291 ) = \imath\hbar\delta^{\scriptscriptstyle{d}-1}(\vec x-\vec y\ , ) \ , .
\label{commutator}\ ] ] combining the equations above we find : @xmath292(\vec x,\vec y;t ) \nonumber\\ & = & ( a_{\mathrm{r}}-c)^{-1}\cdot(a_{\mathrm{r}}+c)(\vec x,\vec y;t ) \ , .
\label{delta function : phi}\end{aligned}\ ] ] this is the desired field theoretic generalisation of the gaussian invariant @xmath293 in equation ( [ delta function:3 ] ) . notice that the result above applies for general non - diagonal gaussian density matrices .
this is an important quantity because , just as in the quantum mechanical case , @xmath126 will be the conserved quantity under any quadratic hamiltonian evolution .
since the von neumann entropy is also conserved in this case , it is natural to expect that @xmath294 $ ] .
of course , if we are interested in problems in which the hamiltonian density is only time dependent , one can make use of spatial translation invariance of the correlators , such that the equal time statistical correlator is homogeneous : @xmath295 . in this case
it is beneficial to fourier transform according to : @xmath296 such that equation ( [ delta function : phi ] ) becomes local in momentum space and reduces to the result known in the literature : @xmath297 \right|_{t = t^\prime } \nonumber\\ & = & \frac{(\tilde a_{\mathrm{r}}+ \tilde c)(\vec k , t ) } { ( \tilde a_{\mathrm{r}}- \tilde c)(\vec k , t ) } \ , .
\label{delta function : phi : momentum}\end{aligned}\ ] ] this representation is particularly useful in problems with spatial translational symmetry , such as cosmology @xcite .
let us now discuss the von neumann entropy ( [ entropy : vn ] ) of the gaussian density matrix ( [ density operator : field ] ) by using the replica method ( [ replica trick ] ) .
one can proceed analogous to section [ gaussian entropy from the replica trick ] .
some subtleties arise however as we deal with a system with infinite degrees of freedom . for this reason
, we nevertheless include an outline of the proof in appendix [ appendix b entropy for gaussian field theory ] .
the entropy for a quantum system that can be described by a gaussian density matrix is given by : @xmath298 \ , .
\label{entropy : vn : delta}\ ] ] we denote the identity matrix by @xmath299 . as in the quantum mechanical case
, we can define the generalised statistical particle number density correlator @xmath300 as : @xmath301 in terms of which the entropy ( [ entropy : vn : delta ] ) reads : @xmath302 \ , .
\label{entropy : vn : n}\ ] ] in the limit when @xmath303 , in the sense that for the diagonalised matrix its diagonal entries of interest are large , we can expand the logarithms in ( [ entropy : vn : n ] ) to get for the entropy , @xmath304 $ ] , which nearly coincides with the wigner entropy @xmath305 in field theory , _ cf .
_ equation ( [ wigner function : entropy:2 ] ) .
equations ( [ entropy : vn : n ] ) and ( [ entropy : vn : delta ] ) represent the von neumann entropy of a general gaussian state in scalar field theory and are the main result of this section . in the homogeneous limit ,
we can again fourier transform and equation ( [ entropy : vn : delta ] ) reduces to : @xmath306 \ , , \label{entropy : vn : delta_fouriertransform}\ ] ] where the volume factor arises because of the trace .
the limit @xmath307 basically agrees with @xcite .
analogous to the one particle non - gaussian entropy discussed in section [ non - gaussian entropy : two examples ] , let us generalise that result to the field theoretical case .
the non - gaussian density matrix @xmath308 in equation ( [ density operator : field ] ) generalises to : [ density operator : field : nong ] @xmath309 where : @xmath310 & = & { \cal n}_2\left[1+\zeta_0\cdot\phi + \zeta_0^\dagger\cdot\phi^\prime + \phi^t\cdot\zeta_1\cdot\phi + { \phi^\prime}^t\cdot\zeta_1^\dagger\cdot\phi^\prime + \phi^t\cdot\zeta_2\cdot{\phi^\prime }
+ { \phi^\prime}^t\cdot\zeta_2^\dagger\cdot\phi \right]\rho_0[\phi,\phi^\prime;t ] \label{density operator : field : nongb } \\
\rho_0[\phi,\phi^\prime;t ] & = & \exp\left[-\phi^t\!\cdot a\cdot\phi - { \phi^\prime}^t\!\cdot a^{\dag}\!\cdot\!\phi^\prime + \phi^t\!\cdot c\!\cdot\!\phi^\prime + { \phi^\prime}^t\!\cdot c^t\cdot\!\phi \right ] \ , , \label{density operator : field : nongc}\end{aligned}\ ] ] we could of course perform a similar shift in the exponent as we have done before in equation ( [ a+c : rescaling ] ) in which case the resulting equation for the entropy would formally be exact , i.e : we do not assume yet that the non - gaussian contributions are small . if we want to expand around the gaussian result however , we can only perform the integral if we assume that all correlators , including the non - gaussian ones , are homogeneous , i.e. : they are only a function of the difference of their coordinates @xmath311 .
the fourier transform of equation ( [ density operator : field : nong ] ) is : @xmath312 = \prod_{\mathbf{k } } { \cal n}_{2,\mathbf{k } } \exp\left[- \bar a(\mathbf{k},t)|\phi(\mathbf{k})|^{2}- \bar a^{\ast}(\mathbf{k},t)|\phi'(\mathbf{k})|^{2 } + 2 \bar c(\mathbf{k},t)\phi^{\ast}(\mathbf{k } ) \phi'(\mathbf{k } ) \right ] + { \cal o}(\zeta_i^2 ) \ , , \label{density operator : field : nong2}\ ] ] where we have set @xmath313 as before , as it will not induce an entropy shift .
also , the product over the momenta is only over half of the fourier space .
moreover , we have absorbed the small non - gaussian contributions in the functions @xmath314 and @xmath315 as in the quantum mechanical case : [ a+c : rescalingqft ] @xmath316 now we can read off the result for the entropy in fourier space in equation ( [ entropy : vn : delta_fouriertransform ] ) as : @xmath317 \ , , \label{entropy : vn : delta_fouriertransformngcase}\ ] ] assuming that @xmath318 we can derive the change in entropy to linear order in @xmath196 .
the result is : [ s2:finalqft ] @xmath319 where @xmath320 is the gaussian contribution and @xmath321 is an entropy shift given by : @xmath322 \label{s2:finalbqft}\,,\end{aligned}\ ] ] where , of course , @xmath323 .
this result is the field theoretic generalisation of the quantum mechanical entropy ( [ s2:final ] ) .
it can be applied to mildly non - gaussian states in field theory , which , apart from being mixed , also contain small one particle contributions .
the field theoretical generalisation of the second example presented in section [ kurtosis ] is hard to solve for , even in the homogeneous case , as it is non - local in fourier space . as a simple illustration of the ideas presented above ,
let us investigate the effect of a changing mass on the gaussian entropy for a scalar field .
this is maybe not a very exciting example , as no entropy is generated of course in free theories , it nevertheless provides an intuitive way of how the wigner function can be used .
let us consider the action of a free scalar field : @xmath324 = \int \mathrm{d}^{4}\!x \left\{-\frac{1}{2 } \partial_\mu\phi(x ) \partial_\nu \phi(x ) \eta^{\mu\nu } - \frac{1}{2 } m^{2}_{\phi}(t ) \phi^{2}(x ) \right\}\,,\ ] ] where as usual @xmath325 is the minkowski metric , and where we consider the following behaviour of the mass @xmath326 of the scalar field , mediated for example by some other higgs - like scalar field : @xmath327 the equation of motion following from ( [ action1 ] ) reads : @xmath328 let us quantise our fields in @xmath329-dimensions by making use of creation and annihilation operators : @xmath330 the annihilation operator acts on the vacuum as usual @xmath331 .
we impose the following commutation relations : @xmath332 = ( 2\pi)^{\scriptscriptstyle{d}-1 } \delta^{\scriptscriptstyle{d}-1}(\vec{k}-\vec{k'})$ ] .
the mode functions @xmath333 of @xmath6 , defined by relation ( [ quantisationphi ] ) thus obey : @xmath334 where @xmath335 and @xmath336 .
using equation ( [ f:1 ] ) with @xmath337 , we see that the mode functions determine the statistical propagator completely : @xmath338 the solution of ( [ eom2 ] ) which behaves as a positive frequency mode in the asymptotic past , i.e. : @xmath339/\sqrt{2\omega_{\mathrm{in}}}$ ] , can be expressed in terms of gauss hypergeometric function @xmath340 ( see @xcite ) : @xmath341 \phantom{1}_{2}f_{1}\left ( 1 + \imath \frac{\omega_{-}}{\rho } , \imath \frac{\omega_{-}}{\rho } ; 1- \imath \frac{\omega_{\mathrm{in}}}{\rho};\frac{1}{2}\{1+\tanh(\rho t)\}\right ) \,,\ ] ] where we defined @xmath342 , @xmath343 and @xmath344 . having the mode functions at our disposal
, we can find the rather cumbersome expressions for the exact statistical propagator .
the statistical propagator in turn fixes the phase space area through equation ( [ delta function : phi : momentum ] ) , yielding @xmath345 such that : @xmath346 .
the squeezed phase space ellipses are plotted at times @xmath347 ( green ) , @xmath348 ( blue ) , @xmath349 ( red ) and @xmath350 ( black ) .
the rotation is clearly visible and its direction is indicated by the arrow .
this can be understood from realising that an arbitrary squeezed state can be interpreted as a superposition of ordinary coherent states and the knowledge that the latter show the same rotating behaviour . [
fig : wigner2 ] ] _ _ .
the squeezed phase space ellipses are plotted at times @xmath347 ( green ) , @xmath348 ( blue ) , @xmath349 ( red ) and @xmath350 ( black ) .
the rotation is clearly visible and its direction is indicated by the arrow .
this can be understood from realising that an arbitrary squeezed state can be interpreted as a superposition of ordinary coherent states and the knowledge that the latter show the same rotating behaviour .
[ fig : wigner2 ] ] _ _ we thus conclude that a changing mass does not change the entropy for a free scalar field .
let us now examine the same process in wigner space . of course , we will reach a similar conclusion ( [ sols2 ] ) but wigner space is much more suited to visualise the process neatly .
the statistical propagator is also the essential building block for the wigner function which can be appreciated from generalising equations ( [ ai : ar : c ] ) and ( [ wigner function : gauss2 ] ) to : [ wignercoefficients ] @xmath351 we can thus plot cross - sections of the 2-dimensional gaussian phase space distribution in the wigner representation . in figure
[ fig : wigner1 ] we depict the squeezing of the gaussian phase space due to the changing mass . at early times , the state is still in an unsqueezed vacuum . at @xmath352
some squeezing is visible , whereas at late times @xmath353 this is manifest . despite the effect of the changing mass on the accessible phase space in wigner space , the area of
the ellipse remains constant throughout the whole process as expected from equation ( [ sols2 ] ) .
at late times , when the mass has settled to its final constant value @xmath354 , the squeezed ellipse rotates in wigner space , which we depict in figure [ fig : wigner2 ] , but it is not squeezed any further .
the rotation can be anticipated by the intuitive notion that an arbitrary squeezed state can be thought of as a superposition of coherent states
( just imagine that the ellipse is replaced by a number of circles displaced from the origin ) .
coherent states that are displaced from the origin rotate in time .
we have formally developed a novel approach to decoherence in quantum field theory : neglecting observationally inaccessible correlators will give rise to an increase in the entropy of the system .
this is inspired by realising that correlators are measured in quantum field theories and that higher order n - point functions are usually perturbatively suppressed .
an important advantage of this approach is that the procedure of renormalisation can systematically be implemented in this framework .
we have shown how knowledge about the correlators of a system affects the notion of the entropy associated to that state in two cases : we firstly calculated the entropy of a general gaussian state .
this entropy can be expressed purely in terms of the three equal time correlators characterising the gaussian state .
moreover , all three correlators can be obtained from the statistical propagator which opens the possibility to study quantum corrections on the entropy in an interacting quantum field theory in a systematic manner .
secondly , we calculated the entropy for two specific types of non - gaussian states .
firstly , we assumed that the state of the system can be described by an admixture of the ground state and a small contribution of the first excited state .
this yields a small correction to the gaussian entropy .
secondly , in the quantum mechanical case , we calculated an expression for the entropy in case our observer could probe a specific type of kurtosis of the ground state .
this also changes the entropy that the observes associates to the state .
we also outlined the use and limitations of the wigner function , the wigner transform of the density matrix of a system .
in particular , we have shown that @xmath126 , the fundamental quantity constructed from various correlators that fixes the entropy , indeed coincides with the phase space area in wigner space .
this is a rather phenomenological discussion , connecting the notion of entropy , correlators and phase space area in quantum field theory .
although we have outlined various mechanisms how entropy could be generated , we have in the present paper not applied our ideas to concrete systems .
we refer the reader to e.g. @xcite for specific examples of how entropy is generated in interacting quantum field theories in an out - of - equilibrium setting .
* jfk and tp thank theo ruijgrok and gratefully acknowledge the financial support from fom grant 07pr2522 and by utrecht university .
the authors also gratefully acknowledge the hospitality of the nordic institute for theoretical physics ( nordita ) during their stay at the `` electroweak phase transition '' workshop in june , 2009 .
here we quote some integrals that are used in section [ non - gaussian entropy : two examples ] .
the three integrals are : @xmath355 \nonumber\\ & & \hskip 5 cm = \ , ( \eta_0 + \eta_0^*)\frac34\frac{\pi^{\frac{1+\epsilon}{2 } } } { ( 2\beta)^\frac{5+\epsilon}{2 } } \frac{(1-\alpha^{2 + 2\epsilon})^2}{(1-\alpha^2)^2(1-\alpha^{1+\epsilon})^5 } \label{app : eta0:int } \\
i_{\eta 1}(\epsilon)&=&\int_{-\infty}^\infty \mathrm{d}x_0 \cdots \int_{-\infty}^\infty \mathrm{d}x_\epsilon \left ( \eta_1 x_\epsilon^3x_0 + \eta_1^ * x_\epsilon x_0 ^ 3 \right ) \exp \left[-2a_{\mathrm{r}}(x_0 ^ 2 + \cdots + x_\epsilon^2)+2c(x_0x_1+\cdots + x_\epsilon x_0 ) \right ] \nonumber\\ & & \hskip 5 cm = \ , ( \eta_1 + \eta_1^*)\frac34\frac{\pi^{\frac{1+\epsilon}{2 } } } { ( 2\beta)^\frac{5+\epsilon}{2 } } \frac{\alpha(1+\alpha^{-1+\epsilon})(1-\alpha^{2 + 2\epsilon } ) } { ( 1-\alpha^2)^2(1-\alpha^{1+\epsilon})^4 } \label{app : eta1:int } \\
i_{\eta 2}(\epsilon)&=&\int_{-\infty}^\infty \mathrm{d}x_0 \cdots \int_{-\infty}^\infty \mathrm{d}x_\epsilon
\left(\eta_2 x_\epsilon^2 x_0 ^ 2 \right ) \exp \left[-2a_{\mathrm{r}}(x_0 ^ 2 + \cdots + x_\epsilon^2)+2c(x_0x_1+\cdots + x_\epsilon x_0 ) \right ] \nonumber\\ & & \hskip 5 cm = \ , \eta_2\frac14\frac{\pi^{\frac{1+\epsilon}{2 } } } { ( 2\beta)^\frac{5+\epsilon}{2 } } \frac{(1+\alpha^2)^2(1+\alpha^{-2 + 2\epsilon } ) -\alpha^4(1+\alpha^{-6 + 2\epsilon})+6\alpha^{1+\epsilon } } { ( 1-\alpha^2)^2(1-\alpha^{1+\epsilon})^3 } \,,\qquad \label{app : eta2:int}\end{aligned}\ ] ] where @xmath356 is an integer . the first non - trivial check of these integrals is to set @xmath357 , in which case their sum gives the ( inverse ) normalisation constant ( [ caln4 ] ) , @xmath358 ^ 2 } = { \cal n}_4^{-1}-{\cal n}_{\rm g}^{-1 } \ , .
\nonumber % \label{app : i : check : eps0}\ ] ] moreover , we encourage the reader to check the integrals for some other values of @xmath149 .
by performing direct integrations we have checked the integrals ( [ app : eta0:int][app : eta2:int ] ) for @xmath359 , and they all give the correct answer . since the expressions ( [ app : eta0:int][app : eta2:int ] ) are analytic in @xmath149 when @xmath360 , they can be uniquely analytically extended to complex @xmath149 s in that range of @xmath159 , based on which we calculate the von neumann entropy in section [ non - gaussian entropy : two examples ] . for the entropy calculation
we need to expand the integrals ( [ app : eta0:int][app : eta2:int ] ) around @xmath357 up to linear order in @xmath149 . from equation ( [ rho4:lin ] ) we see that it is convenient to multiply the integrals with @xmath361^{(1+\epsilon)/2}$ ] , @xmath362 ^ 2 } \bigg\ { 1+\epsilon\bigg [ \ln(1-\alpha ) + \frac{\alpha(5+\alpha)}{1-\alpha^2}\ln(\alpha ) \bigg ] \bigg\ } \label{app : eta0:int : b } \\ { \cal n}_{\rm g}^{1+\epsilon}i_{\eta 1}(\epsilon ) & = & \frac34\frac{\eta_1 + \eta_1^*}{[2(a_{\mathrm{r}}-c)]^2 } \bigg\ { 1+\epsilon\bigg [ \ln(1-\alpha ) + \frac{1 + 3\alpha+2\alpha^2}{1-\alpha^2}\ln(\alpha ) \bigg ] \bigg\ } \label{app : eta1:int : b } \\
{ \cal n}_{\rm g}^{1+\epsilon}i_{\eta 2}(\epsilon ) & = & \frac34\frac{\eta_2}{[2(a_{\mathrm{r}}-c)]^2 } \bigg\ { 1+\epsilon\bigg [ \ln(1-\alpha ) + \frac13\frac{4 + 7\alpha+7\alpha^2}{1-\alpha^2}\ln(\alpha ) \bigg ] \bigg\ } \,,\qquad \label{app : eta2:int : b}\end{aligned}\ ] ] where we used @xmath220 , @xmath363 , and we rescaled @xmath238 by the gaussian width of the state squared , @xmath364 to write it in natural dimensionless units .
here we outline how to calculate the von neumann entropy in field theory for a quantum state that can be described by a general gaussian density matrix . keeping equation ( [ replica trick ] ) in mind
, we realise that , as in the quantum mechanical case , we have to evaluate : @xmath365 & = & { \cal n}^{1+\epsilon}\ ! \int\ ! { \cal d}\phi_0 \cdots\ ! \int { \cal d}\phi_{\epsilon } \exp\big[-2(\phi_0^t\!\cdot a_{\mathrm{r}}\cdot\phi_0 + \cdots + \phi_\epsilon^t\!\cdot a_{\mathrm{r}}\cdot\phi_\epsilon ) \label{replica trick : field:1 } \\ & & \qquad\qquad\qquad\qquad\qquad\qquad\qquad + \phi_0^t\!\cdot c\cdot\phi_1 + \phi_1^t\!\cdot c\cdot\phi_0 + \cdots + \phi_\epsilon^t\!\cdot c\cdot\phi_0 + \phi_0^t\!\cdot c\cdot\phi_\epsilon \big ] \nonumber\\ & = & { \cal n}^{1+\epsilon}\ ! \int\ !
{ \cal d } \breve\phi_0 \cdots\ ! \int { \cal d } \breve\phi_{\epsilon } \exp\big[-2(\breve\phi_0^t\!\cdot \breve{a}_{\mathrm{r } } \mathbb{i } \cdot \breve\phi_0 + \cdots + \breve\phi_\epsilon^t\!\cdot \breve{a}_{\mathrm{r } } \mathbb{i } \cdot \breve\phi_\epsilon ) \nonumber \\ & & \qquad\qquad\qquad\qquad\qquad\qquad\qquad + \breve\phi_0^t\!\cdot \breve{c } \mathbb{i } \cdot \breve \phi_1 + \breve \phi_1^t\!\cdot \breve c \mathbb{i } \cdot \breve \phi_0 + \cdots + \breve\phi_\epsilon^t\!\cdot \breve{c } \mathbb{i } \cdot \breve \phi_0 + \breve \phi_0^t\!\cdot \breve{c } \mathbb{i } \cdot \breve\phi_\epsilon \big ] \nonumber\\ & = & { \cal n}^{1+\epsilon}j\ ! \int\
! { \cal d}\tilde\phi_0 \cdots\ ! \int
{ \cal d}\tilde\phi_{\epsilon } \exp\left[-2(\tilde\phi_0^t\!\cdot \beta \mathbb{i } \cdot\tilde \phi_0 + \cdots + \tilde\phi_\epsilon^t\!\cdot \beta \mathbb{i } \cdot\tilde\phi_\epsilon)\right ] \nonumber \,,\end{aligned}\ ] ] where we note that @xmath366 and @xmath367 are real and symmetric matrices and we assume that they can both be diagonalised by the same orthogonal matrix @xmath368 .
we transformed to the diagonal field coordinates by setting e.g. @xmath369 , such that all matrices diagonalise , e.g. : @xmath370 .
moreover , we assumed that @xmath371 .
secondly , we transformed variables to @xmath372 for @xmath373 and @xmath374 , analogous to the quantum mechanical case .
the jacobian of this change of variables is now given by : @xmath375\right| = { \rm det } \left [ \left(\mathbb{i}-\alpha \mathbb{i } \right)^{-1 } \right]\,,\ ] ] where we write down the matrix @xmath376 for clarity as : @xmath377 the diagonal matrices @xmath378 and @xmath379 generalise to : @xmath380 where we note that @xmath381 . equation ( [ replica trick : field:1 ] ) can now easily be evaluated : @xmath382 = { \rm det}\left[\left ( \frac{(1-\alpha)^{\epsilon+1}}{1-\alpha^{\epsilon+1}}\right)\mathbb{i } \right ] \ , . \label{replica
trick : field:3}\ ] ] expanding around @xmath357 we get : @xmath383 \ , , \label{entropy : vn : alpha : d}\ ] ] where we made use of , @xmath384 = 1 + \epsilon{\rm tr}[a]+{\cal o}(\epsilon^2)$ ] .
the phase space invariant @xmath385 in equation ( [ delta function : phi ] ) can also be cast in diagonal form : @xmath386 we can thus derive @xmath387 .
the resulting equation for the entropy can be rotated back to the original non - diagonal field coordinates by making use of the unitary matrix @xmath368 : @xmath388 \ , , \label{entropy : vn : delta2}\ ] ] which is nothing but equation ( [ entropy : vn : delta ] ) .
999 h. d. zeh , on the interpretation of measurement in quantum theory , found .
* 1 * ( 1970 ) e. joos and h. d. zeh , the emergence of classical properties through interaction with the environment , z. phys .
b * 59 * ( 1985 ) 223 .
paz and w. h. zurek , environment - induced decoherence and the transition from quantum to classical , 77 - 148 in `` fundamentals of quantum information : quantum computation , communication , decoherence and all that , '' d. heiss , editor ( springer , berlin , 2002 ) w. h. zurek , decoherence , einselection , and the quantum origins of the classical , rev .
* 75 * ( 2003 ) 715 .
d. campo and r. parentani , decoherence and entropy of primordial fluctuations .
i : formalism and interpretation , phys . rev .
d * 78 * ( 2008 ) 065044 [ arxiv:0805.0548 [ hep - th ] ] .
e. a. calzetta and b. l. hu , correlation entropy of an interacting quantum field and h - theorem for the o(n ) model , phys .
d * 68 * ( 2003 ) 065027 [ arxiv : hep - ph/0305326 ] .
k. c. chou , z. b. su , b. l. hao and l. yu , equilibrium and nonequilibrium formalisms made unified , phys .
* 118 * ( 1985 ) 1 .
r. d. jordan , effective field equations for expectation values , phys .
d * 33 * ( 1986 ) 444 .
e. calzetta and b. l. hu , nonequilibrium quantum fields : closed time path effective action , wigner function and boltzmann equation , phys .
d * 37 * ( 1988 ) 2878 .
g. aarts and j. berges , nonequilibrium time evolution of the spectral function in quantum field theory , phys .
d * 64 * ( 2001 ) 105010 [ arxiv : hep - ph/0103049 ] .
g. aarts and j. berges , classical aspects of quantum fields far from equilibrium , phys .
* 88 * ( 2002 ) 041603 [ arxiv : hep - ph/0107129 ] .
g. aarts , d. ahrensmeier , r. baier , j. berges and j. serreau , far - from - equilibrium dynamics with broken symmetries from the 2pi-1/n expansion , phys .
d * 66 * ( 2002 ) 045008 [ arxiv : hep - ph/0201308 ] .
j. berges and j. serreau , parametric resonance in quantum field theory , phys .
* 91 * ( 2003 ) 111601 [ arxiv : hep - ph/0208070 ] .
s. juchem , w. cassing and c. greiner , quantum dynamics and thermalization for out - of - equilibrium @xmath389-theory , phys .
d * 69 * ( 2004 ) 025006 [ arxiv : hep - ph/0307353 ] .
s. juchem , w. cassing and c. greiner , nonequilibrium quantum - field dynamics and off - shell transport for phi**4-theory in 2 + 1 dimensions , nucl .
phys . a * 743 * ( 2004 ) 92 [ arxiv : nucl - th/0401046 ] .
j. berges , introduction to nonequilibrium quantum field theory , aip conf .
* 739 * ( 2005 ) 3 [ arxiv : hep - ph/0409233 ] . j. f. koksma , t. prokopec and m. g. schmidt , decoherence in an interacting quantum field theory : the vacuum case , arxiv:0910.5733 [ hep - th ] . j. f. koksma , t. prokopec and m. g. schmidt , in preparation .
a. wehrl , general properties of entropy , rev .
* 50 * ( 1978 ) 221 .
m. c. mackey , the dynamic origin of increasing entropy , rev .
* 61 * ( 1989 ) 981 . s. m. barnett and s. j. d. phoenix , information theory , squeezing , and quantum correlations , phys .
a * 44 * ( 1991 ) 535 .
a. serafini , f. illuminati and s. de siena , von neumann entropy , mutual information and total correlations of gaussian states , j. phys .
b b * 37 * ( 2004 ) l21 [ arxiv : quant - ph/0307073 ] .
s. cacciatori , f. costa and f. piazza , renormalized thermal entropy in field theory , phys .
d * 79 * ( 2009 ) 025006 [ arxiv:0803.4087 [ hep - th ] ] .
g. adesso , entanglement of gaussian states , arxiv : quant - ph/0702069 .
m. hillery , r. f. oconnell , m. o. scully and e. p. wigner , distribution functions in physics : fundamentals , phys .
* 106 * ( 1984 ) 121 .
r. j. glauber , coherent and incoherent states of the radiation field , phys .
rev . * 131 * ( 1963 ) 2766 .
k. e. cahill and r. j. glauber , density operators and quasiprobability distributions , phys .
rev . * 177 * , 1882 ( 1969 ) .
s. mrowczynski and u. w. heinz , towards a relativistic transport theory of nuclear matter , annals phys .
* 229 * ( 1994 ) 1 .
t. prokopec , m. g. schmidt and s. weinstock , transport equations for chiral fermions to order h - bar and electroweak baryogenesis , annals phys .
* 314 * ( 2004 ) 208 [ arxiv : hep - ph/0312110 ] .
t. prokopec , m. g. schmidt and s. weinstock , transport equations for chiral fermions to order h - bar and electroweak baryogenesis .
ii , annals phys .
* 314 * ( 2004 ) 267 [ arxiv : hep - ph/0406140 ] .
r. h. brandenberger , t. prokopec and v. f. mukhanov , the entropy of the gravitational field , phys .
d * 48 * ( 1993 ) 2443 [ arxiv : gr - qc/9208009 ] .
t. prokopec , entropy of the squeezed vacuum , class .
* 10 * ( 1993 ) 2295 .
d. polarski and a. a. starobinsky , semiclassicality and decoherence of cosmological perturbations , class .
* 13 * ( 1996 ) 377 [ arxiv : gr - qc/9504030 ] . c. kiefer , d. polarski and a. a. starobinsky , entropy of gravitons produced in the early universe , phys .
d * 62 * ( 2000 ) 043518 [ arxiv : gr - qc/9910065 ] .
d. campo and r. parentani , inflationary spectra , decoherence , and two - mode coherent states , int . j. theor
. phys .
* 44 * ( 2005 ) 1705 [ arxiv : astro - ph/0404021 ] .
d. campo and r. parentani , inflationary spectra and partially decohered distributions , phys .
d * 72 * ( 2005 ) 045015 [ arxiv : astro - ph/0505379 ] .
t. prokopec and g. i. rigopoulos , decoherence from isocurvature perturbations in inflation , jcap * 0711 * ( 2007 ) 029 [ arxiv : astro - ph/0612067 ] . c. kiefer , i. lohmar , d. polarski and a. a. starobinsky , pointer states for primordial fluctuations in inflationary cosmology , class .
* 24 * ( 2007 ) 1699 [ arxiv : astro - ph/0610700 ] . c. kiefer , i. lohmar , d. polarski and a. a. starobinsky , origin of classical structure in the universe , j. phys .
* 67 * ( 2007 ) 012023 .
d. campo and r. parentani , decoherence and entropy of primordial fluctuations ii . the entropy budget , phys . rev .
d * 78 * ( 2008 ) 065045 [ arxiv:0805.0424 [ hep - th ] ] .
j. f. koksma , decoherence of cosmological perturbations .
a. giraud and j. serreau , decoherence and thermalization of a pure quantum state in quantum field theory , arxiv:0910.2570 [ hep - ph ] .
j. f. koksma , decoherence in quantum field theory , poster and poster presentation at the `` trends in theory '' conference , 14 may 2009 , dalfsen , the netherlands .
j. f. koksma and t. prokopec , quantum reflection in a thermal bath , talk at the `` electroweak phase transition '' workshop , 26 june 2009 , nordita , stockholm , sweden .
m. sohma , a. s. holevo and o. hirota , capacity of quantum gaussian channels , phys .
a * 59 * ( 1999 ) 1820 m. sohma , a. s. holevo and o. hirota , on quantum channel capacity for squeezed states , in quantum communication , computing , and measurement 2 ( springer ) . c. g. .
callan and f. wilczek , on geometric entropy , phys .
b * 333 * ( 1994 ) 55 [ arxiv : hep - th/9401072 ] .
p. calabrese and j. cardy , entanglement entropy and conformal field theory , arxiv:0905.4013 [ cond-mat.stat-mech ] .
p. calabrese and j. l. cardy , entanglement entropy and quantum field theory , j. stat .
mech . * 0406 * ( 2004 ) p002 [ arxiv : hep - th/0405152 ] .
r. kubo , statistical mechanics , north holland , new york ( 1965 ) e. a. calzetta and b. l. hu , thermalization of an interacting quantum field in the ctp-2pi next - to - leading - order large n scheme , arxiv : hep - ph/0205271 . j. f. koksma , t. prokopec and g. i. rigopoulos , the scalar field kernel in cosmological spaces , class .
* 25 * ( 2008 ) 125009 [ arxiv:0712.3685 [ gr - qc ] ] . c. w. bernard and a. duncan , regularization and renormalization of quantum field theory in curved space - time , annals phys .
* 107 * ( 1977 ) 201 .
n. d. birrell and p. c. w. davies , quantum fields in curved space , cambridge monographs on mathematical physics , cambridge university press ( 1982 ) .
|
it is well known that loss of information about a system , for some observer , leads to an increase in entropy as perceived by this observer .
we use this to propose an alternative approach to decoherence in quantum field theory in which the machinery of renormalisation can systematically be implemented : neglecting observationally inaccessible correlators will give rise to an increase in entropy of the system . as an example we calculate the entropy of a general gaussian state and , assuming the observer s ability to probe this information experimentally , we also calculate the correction to the gaussian entropy for two specific non - gaussian states .
|
in this section we briefly introduce the main parts of our experimental setup ( for details , see appendix [ app : setup ] ) , discuss our measurement protocols , and show how we ensure that we measure steady values for the rheology .
we probe the rheology of weakly vibrated granular flows in a modified split - bottom cell , as shown in fig . 1 .
the acrylic container has an inner radius of 7 cm . we mount a hollow smooth cylinder of 6 mm height and 4 cm radius on the bottom of the container .
the rotating disk ( radius @xmath5 of 4 cm and thickness 5 mm ) that drives the granular flow is mounted just above the cylinder .
the gap between the container and the disk is about 0.3 mm so no particles can get underneath the disk . to ensure a no - slip boundary condition ,
the top surface of the disk is made rough by gluing glass particles with diameter of 2 mm to it .
the container is filled with particles ( black soda - lime glass beads , sigmund lindner 4504 - 007-l ) , a polydisperse mixture with a diameter between 1 and 1.3 mm , and a bulk density @xmath6 of 1.7@xmath7 kg / m@xmath8 , up to a filling height @xmath0 . to ensure good reproducibility
, we use to total mass of the particles to control @xmath0 .
all experiments are carried out under ambient temperature , pressure and relative humidity .
we have verified that our experiments are insensitive to relative humidities ranging from 6 and 55% .
after several months of use , the black coating of the particles visibly deteriorates , and the rheological behavior becomes more sensitive to humidity .
we therefore renew our particles on a trimonthly basis , and found that our experiments reproduce well over the course of several years . in the absence of vibrations ,
the phenomenology of the flow is determined by the dimensionless filling height @xmath9 @xcite . in our experiments
, we stay in the low filling height regime ( @xmath10 ) , where the shear bands are mainly vertical , and all grains above the disk co - rotate along with it ( trumpet - flow ) .
we have found that in all but one flow regimes , the flow profiles observed at the free surface are insensitive to the magnitude of vibrations . for the exception which is for slow flow and strong vibrations
see section [ sec : fluid ] . to drive the rotation of the bottom disk and
shear the granular media we use a rheometer ( anton paar dsr 301 ) , which can be used both in stress control ( imposing a torque @xmath2 and measuring the resulting rotation rate @xmath3 ) or in rate control ( impose @xmath3 , measure @xmath2 ) .
we shake the system with a sinusoidal oscillation @xmath11 , with a fixed @xmath12 of 63 hz , using an electromagnetic shaker ( vts systems vg100 ) .
the amount of vibrations is characterized by the dimensionless parameter @xmath1=@xmath13 , where g is the gravitational acceleration .
we image the surface of the system using a mirror and a foculus fo114b camera , allowing us to extract the surface flow by particle image velocimetry . for @xmath3=10@xmath14
rps and @xmath15 but different waiting time @xmath16 between the preshear and the actual measurement . ]
our experiments focus on the rheological curves which relate the driving torque @xmath2 and the driving rate @xmath3 . in section [ sec : constantt ] we describe some experiments performed at constant torque , but our main focus is on experiments where we fix the driving rate in the range from 1 to 10@xmath14 rps , and probe the torque .
we perform these experiments for a range of vibration amplitudes @xmath17 and @xmath18 , and moreover use seven different filling heights ( @xmath19 and @xmath20 ) . varying @xmath21 allows us to probe the role of the confining pressure for the rheology .
each experiments starts with switching on the vibrations , after which we allow the shaker feedback loop 30 s to settle to the required value of @xmath1 .
we proceed by applying an amount of pre - shear to the granular material , in order to obtain similar starting conditions for each experiment . unless noted otherwise , the protocol consists of the following steps : _
( i ) _ 2 s of 1 rps rotation clockwise ; _ ( ii ) _ 4 s of 1 rps rotation counter - clockwise ; _ ( iii ) _ 2 s of 1 rps rotation clockwise .
_ ( iv ) _ 5 s without imposed stress or shear . _
( v ) _ start of actual measurement .
the rotation in the experiments is in the clockwise direction to minimize anisotropy effects @xcite .
vibrations lead to compaction of granular media , although this process is very slow for @xmath22 @xcite , whereas flow typically leads to dilatation @xcite .
additionally , anisotropy in the fabric of the granular media needs a finite amount of strain to build up , but may be relaxed by vibrations @xcite . for both density and anisotropy ,
vibrations and flow are in competition , and as we are interested in steady state flow properties , we need to ask : what is the minimum time or strain necessary to get into a steady state flow regime ? we have probed the relaxation of our flows to a steady state by starting the flow from a denser or less densely packed state as follows : before each experiment , we perform pre - shear as described above .
after pre - shear , during stage _
( iv ) _ , we stop the shear and vibrate the material for a waiting period @xmath16 , during which the granular packing density increases by compaction , and then start the actual measurements . by varying the @xmath16 , we thus manipulate the packing fraction at the start of the flow . by measuring the torque as function of time
we capture the evolution of the torque to its steady state value .
as we expect this equilibration to be slowest for small @xmath3 we perform this test at the smallest @xmath3 ( 10@xmath14 rps ) that we explore in our experiments . .
below @xmath23 , the torque increases with the waiting time .
the measurements are for @xmath24 and @xmath21=0.56 . ]
the results of this test are shown in fig .
[ fig : waiting ] , where we plot @xmath2 as function of the total angle of rotation of the bottom disk @xmath25 .
this figure shows that for small @xmath16 , @xmath2 grows monotonically before reaching steady state , whereas for large @xmath16 , the torque peaks at values larger than the steady state value .
this is consistent with a simple picture where the longer the waiting time , the denser the grains are packed at the start of the experiment , and the larger the torque needed to cause flow . for all waiting times ,
the torque reaches its steady state value for @xmath26 rad , corresponding to a measurement time of 150 s at 10@xmath14 rps .
we claim that this represents the longest equilibration time necessary to reach a steady state flow situation , as all our experiments are carried out for @xmath27 rps .
moreover , in many experiments our data is acquired in a so called strain rate sweep , where the rotation rate is varied by a small amount so that equilibration will be faster . in all cases , an equilibration strain or time of @xmath28 rad , or 150 s , will be sufficient to obtain steady state flow curves .
we choose 180 s for all the experiments described in this paper . to independently verify that equilibration times of 180 s are long enough
, we perform a strain rate sweep at fixed @xmath24 .
we sweep the flow rate from fast to slow rates , and then compare flow curves obtained for different times @xmath29 per step , as shown in fig .
[ fig : aging ] .
while for small values of @xmath3 and @xmath29 , the torque shows a variation with @xmath29 , for all the flow rates probed here we note that the data for @xmath30 s and @xmath31 s are indistinguishable , showing that for @xmath32 s the obtained values of @xmath2 are steady state values . as a final additional test , we have also inspected @xmath33 to confirm we reach steady state ( appendix [ proto ] ) .
= 0.19 ( @xmath34 ) , @xmath21= 0.38 ( @xmath35 ) and @xmath21= 0.56 ( + ) . for @xmath36 and @xmath37
rps , the flow curve always has a negative slope .
the region extends to @xmath38 , and it extends to higher @xmath1 for lower @xmath21 . ] we now turn our attention to the rheological curves @xmath39 . as shown in the @xmath39 curves in fig .
[ fig : overview_g ] , the flows in our system exhibit a rich rheology .
there are two simple trends we see illustrated in these curves : increasing the filling height always increases @xmath2 , whereas increasing the vibration strength always decreases @xmath2 .
the role of the flow rate is not as simple , with the torque often being a non monotonic function of the disk rate @xmath3 moreover , the details of the rheological curves depend on both the vibration strength @xmath1 and filling height @xmath21 .
we note here that the sign of @xmath40 has a crucial rheological implication : flows for which @xmath41 can also be accessed in experiments where the torque is fixed , whereas flows for which @xmath42 are _ unstable _ in torque controlled experiments .
as we discussed in @xcite , this range of unstable flows leads to hysteretic switching between two different flow regimes when the torque is varied , and is deeply connected to the yielding behavior of granular media observed for @xmath36 . here
we focus on rate controlled experiments , and as a first step in characterizing these curves , we plot the boundaries between the regions where @xmath40 is positive and negative for three values of @xmath21 in fig .
[ fig : slope ] . roughly speaking
, we can distinguish three regimes .
_ inertial flows _ for @xmath43 rps , @xmath41 ; the flow curves show an increasing @xmath2 for increasing @xmath3 .
this increase corresponds to the onset of the inertial regime @xcite . to estimate the inertial number @xmath44 at @xmath45 rps
, we have to choose a characteristic pressure and strain rate scale , as both @xmath46 and @xmath47 vary throughout the system .
taking @xmath47 as the hydrostatic pressure at @xmath48 , and @xmath46 corresponding to a shear band of 3 particles wide , we get @xmath49 for @xmath50 and @xmath45 rps .
considering that the inertial regime typically starts at @xmath51 @xcite , there is good agreement between the onset of increasing @xmath39 and the onset of the inertial regime . in the remainder of the paper
we will focus on slower flows
. _ unstable flows _ for intermediate flow rates , @xmath39 has a negative slope for small @xmath1 for @xmath52 , this regime extends to arbitrarily small flow rates , although there the flow curves become essentially flat . despite the unstable character of the global rheology , and in contrast to unstable flows in e.g. micelles @xcite
, we do not see any changes in the shear bands as we move in and out of this unstable regime . as the variation of
the stress with flow rate is not very large , this regime can also be referred to as _
quasistatic_. _ vibration dominated flows _ both the unstable / quasistatic and inertial regime have been studied in great detail already @xcite , as they also arise in the absence of vibrations .
hence , in the remainder of the paper we will focus on the new vibration dominated regime of slow , stable flows that arises for @xmath53 and @xmath54 rps .
one striking qualitative feature of this regime we already want to point out is the pronounced
kink in the flow curves that can be seen for @xmath55 rps@xmath56 rps in fig .
[ fig : overview_g ] . in fig .
[ fig : overview_g]b we explicitly mark such a kink .
the kinks coincide with the flow rates where @xmath29 needs to be sufficiently large for @xmath2 to equilibrate ( see fig .
[ fig : aging ] ) .
we suggest that at sufficiently low @xmath3 , compaction effects become significant , leading to an increase of @xmath2 with time , and a kink in the flow curves .
we will now turn our attention to the increase of @xmath2 with @xmath21 , which allows us to probe the underlying mechanisms that govern the rheology of vibration dominated flows .
the canonical starting point of descriptions of _ non - vibrated _ slow granular flows is that the shear stresses @xmath57 are proportional to the pressure @xmath47 @xcite , and the ratio of @xmath57 and @xmath47 is an effective friction coefficient , @xmath58 . for inertial flows ,
a description where @xmath58 becomes rate dependent ( through the inertial number ) has been shown to capture much of the phenomenology @xcite , and for slow , non - vibrated flows , this mohr - coulomb picture combined with a non - local rheology captures the essentials of steady , slow granular flow @xcite . by varying the filling height @xmath21
, we can modify the pressure @xmath47 and probe its role for the rheology in the different regimes .
clear predictions for @xmath59 exist from a well - studied rheological model for the driving torques in a split - bottom geometry @xcite .
in addition , this model provides clues to the flow s special structure and how it depends on friction and other factors . in this section , we describe how our experiments allow us to build on these basic ingredients to identify two qualitatively different regimes in vibration dominated flows .
we find a frictional regime in which @xmath60 , yet with @xmath61 a rate dependent friction for @xmath62 10@xmath63 rps or @xmath64 .
for even slower , more strongly vibrated flows , both the rheology and the location of the shear band presents strong evidence for a regime where @xmath2 becomes _ independent _ of @xmath47 . as obtained from the fit with the frictional model .
the inset shows one example of the fit for @xmath65 rps .
the upper curve ( + ) shows the raw data @xmath2 , the bottom curve ( @xmath34 ) is the raw data minus the correction term , @xmath66 which goes through the origin .
the fit matches the data very well resulting in a @xmath67 of 2.0@xmath68 ( for the upper curve ) . ] to interpret the observed filling height dependence of the shear stresses , we start from a simple frictional model due to unger _
et al . _ , which was developed to describe the 3d shape of the shear zones in the split - bottom geometry , but which also makes a precise prediction for the driving torque as function of filling height for purely frictional flows @xcite .
this model is based on the following three ingredients .
first , think of the shear zones as localized along a narrow sheet @xmath69 ( corresponding to the center of the shear zones @xcite ) .
second , assume that the stress tensor is colinear with the strain rate tensor @xcite and proportional to the hydrostatic pressure .
third , assume that the sheet shape @xmath69 minimizes the driving torque : , we write the frictional model in terms of @xmath0 , the common notation . ] @xmath70 = 2 \pi g \rho \mu \int^{h}_{0}(h - z)~r^2\sqrt{1+(dr / dz)^2}~dz ~ , \label{ttildeh}\ ] ] where @xmath71 denotes the gravitational acceleration , and @xmath6 the bulk density ( 1.7@xmath7210@xmath8 kg / m@xmath73 ) of the granular material .
minimizing @xmath66 for a given @xmath21 determines the shear sheet @xmath69 , from which the torque can be determined as function of @xmath21 .
as expected , we can write this torque as @xmath74 , where @xmath75 is a universal function of @xmath21 . note that for shallow filling heights , the torque is approximately proportional to the product of pressure and the extension of the shear band , so that @xmath76 is quadratic in @xmath21 for @xmath77 .
in contrast to the original split - bottom cell for which eq .
( [ ttildeh ] ) was developed , in our system the driving disk is slightly elevated with respect to the bottom .
this is done in order to avoid observing spurious torque fluctuations that we associate with the diverging strain rate in the original split - bottom setup .
the elevated disk leads to a @xmath58 depended addition in the experimental torque signal @xmath2 , due to slip between the side of the disk and the stationary particles next to it .
we have found that this drag term can be estimated as : @xmath78 where @xmath79 is the disk thickness ( 5 mm ) , and @xmath58 is the effective friction coefficient for sliding of the disk past the particles , for which we use the same effective friction coefficient as for the granular flow .
the contribution of @xmath80 to the torque varies with @xmath0 and is proportional to @xmath58 , so that we can write @xmath81 , where @xmath82 can be deduced from eq .
( [ drageq ] ) .
we conclude that the measured torque @xmath2 is composed of two contributions : @xmath83=\mu t_f~. \label{ungereq}\ ] ] in conclusion , we can extract @xmath59 from our flow curves , and check whether the flow appears frictional , and if so , determine @xmath58 @xcite and @xmath84 . in fig .
[ fig : overview_h]a we show flow curves for @xmath36 and a range of @xmath21 .
clearly , the torque only weakly varies with @xmath3 , and we expect the stresses to be frictional . for each fixed @xmath3 , we extract @xmath59 from our data and fit it to @xmath85 ( eq . ( [ ungereq ] ) ) , as shown in the inset of fig . [ fig : g0metinset ] .
we find that this fit is excellent , which implies that the stresses are frictional , and which allows us to extract @xmath61 . as shown in fig .
[ fig : g0metinset ] , @xmath61 is almost flat , and has the same shape as the flow curves .
we stress here that @xmath61 together with the frictional model predicts the stresses for _ all _ values of @xmath21 , thus representing all the flow curves taken at different @xmath21 .
we note that our values for @xmath58 are comparable to those found previously in a standard split - bottom cell using the same particles @xcite . in fig .
[ fig : overview_h]b and fig .
[ fig : overview_h]c we show examples of flow curves for a range of @xmath21 and @xmath38 .
we will now use @xmath86 to test if the basic assumptions for the unger model break down in the vibration dominated regime .
we will find two flow regimes with the distinguishing features @xmath87 and @xmath88 .
we describe here how we can distinguish these regimes in the rheological data . as found by fitting the data with eq .
[ ungereq ] .
( b ) the @xmath67 of the fits , as defined in eq .
( [ chi2 ] ) . ] from fig .
[ fig : overview_h]b - c we see that that the flow curves for @xmath38 all show significant rate dependence .
even so , we attempt to fit unger s model to the rheological data .
we thus fit @xmath89 to try to obtain a @xmath61 . if this rate dependence were captured by an effective friction coefficient that only depends on @xmath3 , with @xmath90 , the rate dependence would lead to a good fit of our data to the frictional model . to quantify the deviations between the data and fits to the frictional model
, we calculate the best estimate of @xmath58 and the corresponding @xmath91 as follows . for each fixed @xmath3 and @xmath1 ,
we have measured the torque for seven values of @xmath21 , and then determine : @xmath92 we apply this procedure for each value of @xmath3 and @xmath1 , and show the result for @xmath58 and @xmath67 of these calculations in fig .
[ fig:3x3 ] . for @xmath93 rps - the inertial regime - the fit works very well and results in a weakly rate dependent @xmath58 , just as for @xmath36 .
in addition , we find a large region for @xmath94 and @xmath95 where the fit also works well , but this time with a more strongly rate dependent effective fiction @xmath61 .
this tells us that even in this rate dependent , vibration dominated regime , a frictional prediction is perfectly capable of describing the flow .
we do however observe two distinct regimes where @xmath67 is large , indicating a poor fit .
first , there is a significant peak in @xmath67 around @xmath96 rps for @xmath97 .
second , for @xmath98 and @xmath99 rps , @xmath67 also is substantial .
the underlying physics in these two regimes is different .
as we will show in the next section , the first peak is associated with a broad crossover regime between rate dependent and rate independent flows a rather trivial consequence of the flow profiles in the split - bottom geometry .
the second peak we associate with a flow regime in which the rheology becomes pressure independent , as shown in in section [ sec : fluid ] . for @xmath1=0.95 and a range in @xmath3 at the point where the rate dependence starts .
it can be seen that the curve drop for high @xmath21 , resulting in an s - shaped @xmath59 curve rather than an upwards curved @xmath100 one .
the black line is the fit with the frictional model to the top curve .
( b ) a theoretical prediction of @xmath101 -which decreases with @xmath102- in the split - bottom cell @xcite .
dark color presents high @xmath101 , white is low @xmath101 . ]
the peak in @xmath67 around @xmath96 rps is consistent with the onset of rate dependence below @xmath103 rps as per the following reasoning .
first , both our raw data for @xmath2 as well as the best fits for @xmath58 show that rate dependence sets in rather abruptly for @xmath104 rps , and that rate dependence is strongest for large @xmath1 , consistent with the location and strength of the peak in @xmath67 .
crucially , this onset of rate dependence sets in at different flow rates for different heights ( see fig . [
fig : overview_h ] ) , so that at a given @xmath3 , the data for @xmath59 mixes rate independent and rate dependent flows . in fig .
[ fig : ratedep]a we show examples of @xmath2 as function of height , that illustrate that when @xmath3 enters this rate dependent regime , @xmath59 strongly deviates from the quadratic form predicted by eq .
( [ ungereq ] ) . to interpret this deviation ,
it is important to realize that at a given @xmath3 , the local strain rate @xmath105 spans a wide range of values and has a strong @xmath102 dependence @xcite see fig .
[ fig : ratedep]b . hence , as the torque @xmath2 is an integral over the local stress in different layers in the material , @xmath39 mixes different local rheologies . more precisely
: under the assumption that rate dependence sets in below a given @xmath105 , there is a range of values of @xmath3 for which the lower part of the system ( where strain rates are largest ) is still rate independent , whereas the top part of the system ( where strain rates are smallest ) are already rate dependent .
this is consistent with the drop in the @xmath59 curves at large @xmath21 shown in fig .
[ fig : ratedep]a the deviations from the unger model emerge first for large @xmath21 , for which the range of strain rates is biggest and regime mixing is thus most pronounced .
our data also shows that once @xmath3 is sufficiently low , so that all of the material is in a rate dependent state , @xmath89 is close to @xmath106 so that @xmath67 drops to low values again , and @xmath58 can be replaced with a rate dependent @xmath61 .
the range of @xmath3 over which this crossover exists broadens with @xmath1 , since the rate dependence becomes stronger with @xmath1 . in conclusion ,
the lowering of the friction coefficient @xmath58 and the peak in @xmath67 around @xmath107 rps are caused by the onset of rate dependence which occurs at different @xmath3 for different vertical locations in the flow cell .
for @xmath64 , we also observe that once all the material is in the slow , rate dependent regime , the fit to the frictional model achieves a low @xmath67 again , so that @xmath108 . curves for @xmath109 and @xmath99 rps . towards lower @xmath3 ,
the curves lose their curvature and become straight lines . ]
the growth of @xmath67 for large @xmath1 and low @xmath3 signals a breakdown of the frictional picture , where shear stresses are proportional to the pressure , as we will describe in this subsection . to gain deeper insight in the flow phenomenology in this regime , we plot @xmath59 for @xmath109 and a range in @xmath3 in fig .
[ fig : toline ] .
we see that for all filling heights the stresses drop with @xmath3 , and at low @xmath3 , @xmath59 becomes approximately _
linear_. the standard unger model predicts a quadratic dependence of @xmath59 on @xmath21 , as mentioned above .
a linear dependence would suggest a pressure independent rheology , for which the increase of @xmath2 with @xmath21 is only due to increasing surface area on which the shear stress acts .
note that the large values of @xmath67 here can not be due to the existence of a crossover regime , as presented above for @xmath110 rps .
evidence for this comes from fig .
[ fig : overview_h]c , which shows that both rate dependence of @xmath39 is small , and that there is no strong difference in the rate dependence for different values of the height in this regime .
first of all , that means that there is little mixing of different rheologies in the global torque signal ; second , the rate dependence is weak , so even if there were some mixing , it would not produce a strong @xmath21 dependence .
it it perhaps not surprising that new phenomena occur around the special value @xmath109 . for @xmath111 , the grains
lose contact during part of the vibration cycle the precise value of @xmath1 where this happens depends on details @xcite . as a result ,
the confining pressure becomes zero during part of the cycle , and as most slip can be expected to occur when the normal grain forces are absent , the flows may become pressure independent , as in a viscous liquid . for @xmath112 rps , @xmath113 ( @xmath72 ) , @xmath114 rps , @xmath109 ( + ) , @xmath115 rps , @xmath109 ( @xmath34 ) .
in red , we add the fit with @xmath116 erf@xmath117 $ ] .
( b ) the center of the shear band at the surface @xmath118 as a function of @xmath1 for h=0.47 and @xmath3 as indicated . for low @xmath3 , @xmath118 clearly decreases . ] additional evidence for the loss of pressure dependence for high @xmath1 and low @xmath3 comes from measurements of the flow structure . from finite element calculations on the flow structure of a viscous liquid in the split - bottom geometry
, it is known that the shear band is much closer to the center of the cell than for frictional flow @xcite . as such
, a pressure independent rheology for the granular flows in this regime can be expected to be accompanied by similar changes in the flow structure . to test this ,
we have measured the velocity profiles @xmath119 at the surface of our system for a range in @xmath1 and @xmath3 using particle image velocimetry @xcite . in fig . [
fig : flowprofile]a we show examples of @xmath119 , showing a broadening and shift of the shear zones when @xmath3 enters the pressure independent regime .
we fit the velocity profiles with @xmath116 erf@xmath120 $ ] , where @xmath118 is the center of the shear band at the free surface @xcite . in fig .
[ fig : flowprofile]b , we plot @xmath118 for @xmath3 ranging from @xmath121 to @xmath122 rps and a range in @xmath1 .
clearly , the location of the shear band is mostly independent of @xmath3 and @xmath1 , including in most of the rate dependent regime .
however , in the regime where we observed the pressure independent rheology , we observe significant deviations in the flow profiles .
the deviations show a trend towards a shear band moving inwards consistent with the idea of a pressure independent regime . moreover
, we can modify the unger model to test which rheological scenario is most compatible with the observed shift in the shear band . throughout , we assume that the torque minimization principle is robust .
the frictional torque model assumes @xmath123 , in which the shear stress , being proportional to the hydrostatic pressure , goes to zero at the surface .
we can replace this model with @xmath124 , in which the shear stress reaches a final value when approaching the free surface
see fig .
[ fig : ungershift]a .
the extreme case @xmath125 represents a pressure independent rheology .
we compute the location of the shear band at the free surface as a function of model parameter @xmath126 .
the results are shown in fig .
[ fig : ungershift]b .
we find that for larger @xmath126 , the location of the shear band at the free surface moves inwards .
thus , the closer the model resembles a newtonian rheology , the more the shear band moves towards the center .
this can be understood intuitively as follows : the penalty for having a shear band at large radius at the surface is zero in the pressure dependent model , because the shear stress goes to zero at the free surface .
once a finite amount of shear stress is present in the shear band at the surface , torque minimization will move the shear band inwards precisely as we observe in the experiments at @xmath127 rps , @xmath128 .
conversely , for a frictional , rate dependent rheology , the shear stress closer to the surface is _ lower _ than that of a simple frictional model .
we model this with a @xmath129 that can be captured with @xmath130 , as shown in fig .
[ fig : ungershift]c . the torque penalty for having a shear band at finite @xmath131
is thus reduced , and the model predicts indeed an _ increase _ of the shear band radius at the free surface ( fig . [
fig : ungershift]d ) , contrary to what we observe .
we thus conclude that our observation of the inward displacement of the shear band location at @xmath127 rps , @xmath128 is consistent with the granular flow obtaining a rheology which becomes pressure independent . , can be found using the method by unger for given @xmath102 dependent stress @xmath129 . in ( a )
, we plot @xmath129 for case @xmath124 where @xmath132 is still finite at @xmath133 , in contrast to the frictional description where @xmath134 .
the resulting @xmath118 is shown in ( b ) , we recover in inwards moving shear band as we observe in experiments . in ( c )
we plot @xmath129 for @xmath130 , where @xmath47 and @xmath58 respectively vanish and decrease towards the surface , corresponding the a strain rate dependent frictional picture , as we show in ( d ) , this predict an outwards moving shear band , contrary to what we observe . ]
in this section we discuss stress controlled experiments .
these experiments consist of the following protocol : after pre - shearing the sample , a fixed value of the driving torque @xmath2 is set , and the ensuing rotation angle @xmath25 of the disk is monitored .
we show several examples of the time evolution of @xmath25 and its derivative @xmath3 for a range of values of the torque @xmath2 and vibration strength @xmath1 and fixed @xmath135 in fig .
[ fig : tmerge ] . data for other values of @xmath2 , @xmath1 and @xmath21 look similar ; the small initial oscillatory motion , visible for small @xmath1 , is due to transient flexure oscillations . as shown in fig .
[ fig : tmerge ] , in most cases the disk rotation rate reaches a steady state value after some equilibration time . for small @xmath2 ,
this equilibration time becomes extremely long , much longer than in the rate controlled experiments shown above .
this is typical for complex fluids that exhibit a so - called viscosity bifurcation @xcite , and we suggest that weakly vibrated granular media fall into this category .
we also note that for small @xmath2 , the angle as function of time ( or alternatively , @xmath136 ) exhibits power law behavior reminiscent of andrade creep @xcite , again a typical feature of weakly driven disordered media . for the data sets where @xmath3 reaches a steady state , we find that @xmath137 and @xmath39 are consistent in other words , the steady state flow properties are the same as measured through the rheology under constant rate or constant stress driving .
consistent with this picture , we observe that the data in fig .
[ fig : tmerge ] illustrates that the steady state values of @xmath3 exhibit a jump between inertial rates ( @xmath138 rps ) and significantly slower flow rate ( @xmath54 rps ) , consistent with the hysteretic transition between inertial and vibration dominated flows observed in @xcite . . in ( a )
, we show the @xmath25 as a function of time , in ( b ) , we show the derivative of @xmath25 : @xmath3 . it can be seen that the curves for low @xmath1 and @xmath57 show very long transients .
the black line has a slope -1 .
the curves are down up for @xmath1390.2,0.3,0.4,0.4,0.4,0.5,0.6,0.7,0.75,0.8,0.8,0.8,0.9,0.9,1 and @xmath2 [ mnm]=3.36 , 4.62 , 5.74 , 5.88 , 7.14 , 8.54 , 9.94 , 10.64 , 11.3 , 10.64 , 11.3 , 12.04,13.02,13.44 , as also indicated in the inset of ( b ) . ]
to summarize , we probe the rheology of weakly vibrated granular media and find several different flow regimes .
first , for @xmath93 rps , our data shows the well known _ inertial flow _ regime , consistent with a rough estimate of the inertial number . in the absence of vibrations ,
lower flow rates lead to an essentially rate independent , _ quasistatic _ ,
regime , where the variation of the torque is small , and where @xmath59 is well fitted using eq .
( [ ungereq ] ) , implying that the shear stresses are proportional to the pressure here . for @xmath36 ,
this regime covers all @xmath140 rps , whereas the range of flow rates where this rate independent regime resides shrinks in the presence of vibrations , and almost vanishes for @xmath109 . for @xmath3 below the rate independent regime and @xmath38 ,
we have described three _ vibration dominated _ regimes . for two of these regimes ,
our data shows that the shear stresses are still proportional to the normal stresses , but now via rate dependent @xmath61 . for the slowest of these two regimes ,
we see a slow densification , leading to a kink in the flow curves .
finally , for @xmath1 close to one , the vibrations affect the rheology of the granular medium so significantly , that the shear stresses are no longer proportional to the normal stresses , signifying a complete departure of the frictional nature that is a hallmark of all other types of slow granular flows .
we appreciate helpful discussions with o. dauchot and h. jaeger , j. mesman for outstanding support in constructing the setup , and t. tampung for support in developing the electronics .
we thank l. van dellen for his contribution to the experiments and data analysis . this work was supported by the dutch physics foundation fom .
our setup allows to do sensitive rheology on vibrated materials and its main parts are illustrated in fig .
[ fig : setup ] . the granular split - bottom flow cell forms the heart of setup , and consists of a cylindrical flow cell , the bottom of which is formed by a rotating disk that drives the flow .
the disk is attached to an axle which is mounted in a cylindrical air bearing ( @xmath141 1/4 " , new way ) : rotational motion is friction - free which allows precise rheological measurements . in ordinary split - bottom cells ,
the strain rate diverges at the split .
we have found that the torques occasionally show strong fluctuations in this case , presumable due to individual particle being trapped just above the split moreover , these fluctuations depend on the precise roughness near the split , thus leading to a dependence of the average @xmath2 on such experimental details . to avoid this ,
we make sure that the strain rate field is smooth at the grain level and have elevated the driving disk by 6 mm ( fig .
[ fig : diskfric ] ) .
the side of the disk is smooth , and particles immediately next to the disk hardly move , creating a static bottom layer flush with the disk .
hence , the boundary conditions are essentially the same as for the ordinary split - bottom disk , and the elevation does not affect the overall flow field for @xmath9 larger than about 0.1 .
the elevation does ensue that the torques are insensitive to experimental details and do not show the aforementioned spurious fluctuations .
we induce vibrations in the split - bottom cell with an electromagnetic shaker ( vts systems vg100 ) , driven by a function generator ( thurlby thandar tg1010 ) via a commercial audio amplifier ( crown ce1000 ) .
the shaker is coupled to the flow cell via a long rod , and the cell is mounted on a square air bearing ( 4``x4 '' , new way ) , which ensures that the slider and flow cell move virtually friction free in the vertical direction , with the amplitude of the residual horizontal vibrations in the submicron range due to the stiffness of the air bearing .
the total combined weight of the slider and the split - bottom cell is approximately 12 kg and too large to be carried by the shaker alone .
therefore we support their weight with four suspension springs with a stiffness of 2 kn / m , connected to the aluminum frame shown in fig .
[ fig : setup]a , b .
the vertical acceleration of the split - bottom cell is measured by a combination of two accelerometers ( dytran e3120ak and a modified adxl322eb ) . to limit the mechanical coupling between the table and the shaker , the shaker is placed on a stack of concrete tiles with rubber mats sandwiched between them .
we do not observe any appreciable heaping @xcite .
the vibrations in the flow cell should not induce relative vertical motion between the disk that drives the granular flow and the container . to ensure that the disk only induces rotational shear deformations and no additional vibrations
, we attach a flange to the disks axle .
the surface of the flange is polished , and clamped between four flat air bearings ( @xmath141 25 mm , new way )
see fig .
[ fig : cusairbearing ] . by mounting the flat air - bearing assembly on the container ,
we ensure that disk and container move together during the vibration cycle : the stiffness of one flat bearing , for a typical gap between the bearing and the flange of 5 @xmath58 m , is 18 n/@xmath58 m . since the peak acceleration of the system is never larger than @xmath142 @xmath71 , and the weight of the flange / axle / disk construction is less than a kilogram , the maximum residual motion between the disk and the container is smaller than a micrometer , which is the typical length scale of asperities on glass beads @xcite .
a rheometer ( anton paar dsr 301 ) , coupled to the flow cell via a system of flexures and airbearings , allows to probe and control the granular flow .
as the rheometer can not be vibrated , and its rotational axis is housed in a very stiff air - bearing , we need to decouple the vibrations from the rheometer , and also compensate for inevitable misalignments of rheometer and flow cell .
we therefore couple the two axes with a flexure , as shown in fig .
[ fig : setup]c .
this flexure has a torsional spring constant of 4 nm / rad and compressional spring constant of 5 @xmath143 n / m .
the flexure can be compressed easily and therefore accommodate the vertical motion of the disk axle .
the copper blades used in the flexure are flexible enough to compensate for any small misalignment between the axis of the disk and the rheometer , without exerting significant forces on the bearings .
our dsr 301 rheometer is a stress driven rheometer , so its native mode is to set a torque to control rotation .
it is also possible to use the rheometer in a so - called constant shear rate mode . in this mode
, the rheometer applies a feedback routine that dynamically adjusts its torque in order to achieve a constant rotation rate .
this process has one control parameter : the rate at which the torque is increased or decreased .
this timescale is set by the desired driving rate , and by a user - defined time constant , and in principle could affect all constant shear rate experiments . in these experiments we set the feedback constant such that we observe smooth rotation , which is achieved for a feedback parameter , the so called csr - value , of 25% .
if the torque adaptation rate is too high and becomes of the order of the eigenfrequency of the rheometer axle , the feedback routine will be affected by the rotational eigenmodes of the mass - spring system of the rheometer axle . to compensate for inevitable misalignments between the shaker axis and vertical air - bearing axis , we couple them via a 40 cm long and flexible aluminum rod .
the aluminum rod is flexible enough to compensate for a horizontal alignment mismatch of a few millimeter , but stiff enough to ensure a mechanical coupling for vertical motion ( see fig .
[ fig : setup]a ) .
misalignment between the disk and rheometer axle causes rotationally periodic variations in the force that the rheometer exerts on the disk and grains .
this periodic modulation of the torque can be reduced by improving the alignment of the flexure - coupled axles .
we manually align the rheometer with the disk axis by setting a constant rotation rate to the disk , with no particles in the container . while the disk rotates ,
we monitor the required torque to run the disk at the set speed .
this torque has a periodic modulation ; alignment of the two axis with the set screws on the air bearing reduces the amplitude of this periodic modulation on the torque .
a typical alignment run is shown in fig .
[ fig : residue ] : the modulations can be minimized to an amplitude of less than 20 @xmath58 nm , which is @xmath144 0.1% of the largest torque need to drive our granular flows .
nota that a small torque offset also remains since the flat air bearings that confine the vertical motion of the flange can not be positioned perfectly parallel to the flange .
this induces an asymmetry in the air flow between the bearings and the flange .
this exerts a torque of the order of 20 @xmath58 nm on the flange ; a negligible amount in comparison to the typical driving torques employed in our experiments . to choose an appropriate external vibration frequency
, we measure how the steady state rotation rate @xmath3 under a constant applied torque @xmath146 mnm and fixed vibration amplitude @xmath147 depends on the vibration frequency @xmath12 .
we carry out the measurements with a filling height of @xmath21 = 0.55 .
the result is shown in fig .
[ fig : vibrheo : mechresp ] .
we see that there is a resonance above 105 hz , and for vibration frequencies between 50 and 75 hz , the rotation rate depends only marginally on the vibration frequency . we choose the middle of this regime , @xmath12 = 63 hz to carry out all experiments discussed in this paper .
the vertical vibration amplitude at a peak acceleration of 1 @xmath71 at this frequency is 64 @xmath58 m .
it is critical that the vibration amplitude is constant during the duration of the experiments , which can last for hours .
the shaker generates considerable heat , which affects its operation ; with constant driving current to the shaker coils , the vibration amplitude slowly drifts over @xmath148 per hour . to keep the vibration amplitude constant during the experiments , we employ a labview driven feedback scheme that continuously adjusts the driving current for the shaker coil .
a function generator , under computer control , sends a sine wave to the amplifier that drives our shaker .
accelerometers detect the actual shaking amplitude .
we find that the harmonic distortion of the sinusoidal shaker motion is about 1% , so we use a lock - in amplifier ( srs 830 ) to achieve a high accuracy measurement of the vibration amplitude . the computer then adjust the output of the function generator , and with this feedback scheme we can maintain a stability of @xmath1 to within @xmath149 @xmath71 .
we verify that a steady state flow rate can be obtained by the rheometer and that we observe no long term transients in the rate control mode .
to do this , we measure the instantaneous driving torque and speed as a function of time for different rotation rates @xmath3 and vibration amplitudes @xmath1 .
we apply pre - shear before the start of the experiment , and start a logarithmic ramp down in @xmath3 , from @xmath151 to @xmath152 rps with 2 steps per decade .
we measure for all @xmath3 over a minimum strain of 0.5 rotation , or a total shear time of 10 seconds , whichever is longer .
we sample the rotation rate and torque with a rate of 10 hz . the results for @xmath21 = 0.33 are shown in fig .
[ fig : transient ] , both the instantaneous speed of the disk and the driving torque as a function of time .
note the logarithmic scale on the time axis .
we see that for @xmath153 , the set speeds are achieved very accurately and no large fluctuations or transients are observable .
the fluctuations in the torque signal are however appreciable .
transients are short , even for the slowest run for more on transients , see the next section . for @xmath150 ,
the fluctuations in both speed and torque are strongly reduced and we can clearly see that no long time transients are present in these experiments . we conclude that the mean driving torque @xmath2 and mean rotation rate @xmath3 are well defined quantities in steady state shear experiments .
|
we probe the rheology of weakly vibrated granular flows as function of flow rate , vibration strength and pressure by performing experiments in a vertically vibrated split - bottom shear cell . for slow flows ,
we establish the existence of a novel vibration dominated granular flow regime , where the driving stresses smoothly vanish as the driving rate is diminished .
we distinguish three qualitatively different vibration dominated rheologies , most strikingly a regime where the shear stresses no longer are proportional to the pressure .
granular media are collections of macroscopic , athermal grains which interact through dissipative , frictional contact forces . in the presence of gravity and in the absence of additional external forces , they jam in metastable configurations ; however , external forcing can easily lead to yielding and flow @xcite .
the best known scenario that leads to granular flow is by exerting shear stresses that exceed the yield stress , as when tilting a quiescent layer of sand sufficiently far away from the horizontal @xcite . to understand such dense granular flows
, it is becoming increasingly clear that both stress and mechanical agitations play a crucial role .
indeed , a given stress can give rise to a wide range of flow rates depending on the magnitude of these agitations @xcite .
moreover , agitations make granular media lose their rigidity , although in the absence of shear stresses this does not need to cause flow @xcite .
we note here that the idea that both the stress and the amount of agitations determine the flow rate lies at the basis of numerous models for slowly flowing disordered materials @xcite .
is used to probe the rheology of agitated granular media .
the crucial experimental parameters are the filling height , @xmath0 , vibration amplitude , @xmath1 , the torque , @xmath2 , and the rotation rate , @xmath3.,width=188 ] in a granular context , such agitations may be provided by external vibrations . in a classic experiment ,
the slope of a granular pile was found to relax under vertical vibrations @xcite , and similarly , horizontal vibrations have been used to induce flow on inclined planes @xcite .
piezo transducers inside the medium have been used to inject tiny rearrangements or force fluctuations , either locally @xcite or along a complete boundary @xcite .
shear induced agitations similarly have induced microscopic rearrangements @xcite .
naturally , flow itself also induces mechanical agitations @xcite .
for example , for shear banded flows @xcite , the observation of particle rearrangements and fluidization far away from the flowing region suggest that the effect of agitations can be carried far through the material @xcite .
agitations thus form a crucial ingredient for non - local extensions of models for the rheology of granular flows @xcite , and may explain , among other things , the large extension of shear bands in split - bottom granular flows @xcite . recently
, we have explored how weak vibrations influence the rheology of dry granular media @xcite by performing experiments in a vibrated split - bottom cell , as shown in fig .
[ fig : intro ] .
we found that weak external vibrations suppress the yield stress of the material and strongly influence the rheology of slow granular flows . in this paper
, we reveal the intriguing rheology of weakly vibrated granular media in much more detail .
we find that we can distinguish a variety of qualitatively different flow regimes .
first , for large flow rates , inertial effects dominate , and the effect of vibrations is small .
second , for slower flow rates , we cross over to a regime similar to the well - known quasi static flows that have been studied at length in the absence of vibrations @xcite .
third , for slower flows , we enter a regime where the vibrations lead to completely new rheological behavior .
the focus of this paper is on these _ vibration dominated flows_. by probing the equilibration times of the stresses and the variation of the steady state stresses with filling height , we find evidence for three qualitatively different regimes .
for slow enough flows , vibration effects increasingly dominate the physics , leading to compaction of the material for weak vibrations , and to fluidization of the material for vibrational accelerations approaching gravity .
most strikingly , in the latter regime , we see a breakdown of the proportionality of shear stresses and pressure , a highly unusual phenomenon in granular flows .
we also study how the rheology of weakly vibrated granular media behaves in stress controlled flow experiments .
we find that equilibration times can be dramatically longer than in rate controlled experiments , akin to what has been observed in several soft materials that exhibit a so - called viscosity bifurcation @xcite , and leading to andrade creep like phenomena @xcite , providing an interesting analogy with thermal , disordered flows @xcite .
the outline of this paper is as follows . in section [ sec : setup ]
we describe the details of our experimental setup and the measurement protocols used in the current and previous study @xcite . in section [ sec : pheno ] we describe the main phenomenology of a complete set of experiments probing @xmath4 . in section [ sec : unger ] we introduce the canonical perspectives on granular rheology , including a model for the stresses in split - bottom flows by unger _
et al .
_ @xcite . in section [ sec : ggt0 ] we use this model to extract effective friction coefficients from our data , as well as exploring the quality of the fit between the data and this model . in section
[ sec : fluid ] we provide strong evidence for the existence of a pressure independent flow regime though measurements of the flow structure .
additionally , we compare steady state results obtained at constant @xmath3 driving with a constant stress driving mode in section [ sec : constantt ] .
|
neuronal networks exhibit diverse patters of activity , including oscillations , synchronization and waves . during neuronal activity ,
each neuron can receive inputs by thousands of other neurons and , when it reaches a threshold , redistributes this integrated activity back to the neuronal network .
recently a neuronal activity based on avalanches has been observed in organotypic cultures from coronal slices of rat cortex @xcite where neuronal avalanches are stable for many hours @xcite .
more precisely , recording spontaneous local potentials continuously by a multielectrode array , has shown that activity initiated at one electrode might spread to other electrodes not necessarily contiguous , as in a wave - like propagation .
cortical slices are then found to exhibit a new form of activity , producing several avalanches per hour of different duration , in which non - synchronous activity is spread over space and time . by analysing the size and duration of neuronal avalanches ,
the probability distribution reveals a power law behaviour , suggesting that the cortical network operates in a critical state .
the experimental data indicate for the avalanche size distribution a slope varying between -1.2 and -1.9 , depending on the accuracy of the time - binning procedure , with a value -1.5 for optimal experimental conditions .
interestingly , the power law behaviour is destroyed when the excitability of the system is increased , contrary to what expected since the incidence of large avalanches should decrease the power law exponent .
the distribution then becomes bimodal , i.e. dominated either by very small or very large avalanches as in epileptic tissue .
the power law behaviour is therefore the indication of an optimal excitability in the system spontaneous activity .
moreover the avalanche time duration is also found to follow a power law behaviour as function of the duration time normalised by the binning time with an exponent equal to -2.0 followed by an exponential cutoff .
these results have been interpreted relating spontaneous activity in a cortical network to a critical branching process @xcite , indeed the experimental branching parameter is very close to the critical value equal to one , at which avalanches at all scales exist . on the other hand ,
the dynamics observed in spontaneous brain activity is very similar to self - organized criticality ( soc ) @xcite .
the term soc usually refers to a mechanism of slow energy accumulation and fast energy redistribution driving the system toward a critical state , where the distribution of avalanche sizes is a power law obtained without fine tuning : no tunable parameter is present in the model .
the simplicity of the mechanism at the basis of soc has suggested that many physical and biological phenomena characterized by power laws in the size distribution , represent natural realizations of soc .
for instance , soc has been proposed to model earthquakes @xcite , the evolution of biological systems @xcite , solar flare occurrence @xcite , fluctuations in confined plasma @xcite , snow avalanches@xcite and rain fall @xcite .
moreover , power law behaviour is observed in power spectra of different time series monitoring neural activities .
prominent examples are eeg data which are used by neurologists to discern sleep phases , diagnose epilepsy and other seizure disorders as well as brain damage and disease @xcite .
another example of a physiological function which can be monitored by time series analysis is the human gait which is controlled by the brain @xcite .
for all these time series the power spectrum , i.e. the square of the amplitude of the fourier transformation double logarithmically plotted against frequency , generally features a power law at least over one or two orders of magnitude with exponents between 1 and 0.7 .
moreover , experimental results show that the neurotransmitter secretion rate exhibits fluctuations with time having power law behaviour @xcite and power laws are observed in fluctuations of extended excitable systems driven by stochastic fluctuations @xcite . on the basis of these observations ,
recently a model based on soc ideas and taking into account synaptic plasticity in a neural network @xcite has been proposed .
plasticity is one of the most astonishing properties of the brain , occuring mostly during development and learning @xcite , and can be defined as the ability to modify the structural and functional properties of synapses , properties which are thought to underlie memory and learning . among the postulated mechanisms of synaptic plasticity ,
the activity dependent hebbian plasticity constitutes the most fully developed and influential model of how information is stored in neural circuits @xcite . within a soc approach
the four most important ingredients for neuronal activity have been introduced , namely threshold firing , neuron refractory period , activity - dependent synaptic plasticity and pruning .
the system consists in an electrical network on a square lattice , on which each site represents the cell body of a neuron , each bond a synapse .
therefore , each site is characterized by a potential and each bond by a conductance .
whenever at a given time the value of the potential at a site is above a certain threshold , approximately equal to @xmath0 for the real brain , the neuron fires , i.e. generates an `` action potential '' , distributing charges to its connected neighbours in proportion to the current flowing through each bond . after firing
, a neuron goes back to the resting potential of @xmath1 and remains inactive during the refractory period , when it is unable to send or receive information from other neurons .
this time corresponds for real neurons to the physiological time needed to reset ion channels after the transmission of the action potential through the axon .
the conductances , on the other hand , represent hebbian synapses , for which the conjunction of activity at the presynaptic and postsynaptic neuron modulates the efficiency of the synapse @xcite . to this extent , each time a synapse transmits an action potential between active neurons , its strength is increased proportionally to the intensity of the transmitted signal , whereas synapses inactive during a neuronal avalanche have their strength decreased , as for hebbian rules .
synapses successively weakened may have their strength finally set to zero , i.e. are `` pruned '' , eliminating that particular connection between neurons .
pruning implies that , as activity goes on , the initial regular lattice is transformed , some patterns are strengthened and the connectivity of some neurons decreased .
the system exhibits an avalanche activity power law distributed with an exponent close to -1.5 , as measured for spontaneous activity @xcite .
the analysis of the power spectra of the electrical signal reproduces very robustly the power law behaviour with the exponent 0.8 , experimentally measured in eeg spectra .
the same value of both exponents is found considering leaky neurons or introducing a small precentage of inhibitory synapses , indicating that universality holds for a wide class of brain models . in real brain neurons
are known to be able to develop an extremely high number of connections with other neurons , that is a single cell body may receive inputs from even a hundred thousand presynaptic neurons . one of the most fascinating questions is how an ensemble of living neurons self - organizes , developing connections to give origin to a highly complex system .
the dynamics underlying this process should be driven both by the aim of realizing a well connected network leading to efficient information transmission , and the energetic cost of extablishing very long connections .
the morphological characterization of a neuronal network grown _ in vitro _ has been studied @xcite by monitoring the development of neurites in an ensemble of few hundred neurons from the frontal ganglion of adult locusts .
after few days the cultured neurons have developed an elaborated network with hundreds of connections , whose morphology and topology has been analysed by mapping it onto a connected graph .
the short path length and the high clustering coefficient measured indicate that the network belongs to the category of small - world networks @xcite , interpolating between regular and random networks .
however , the system grown _ in vitro _ necessarily lacks some features of _ in vivo _ systems , therefore the average node connectivity is found equal only to few units and the `` scale - free '' feature @xcite of many real networks was not recovered .
small - world networks are characterized by an efficient information transmission with a small number of long range connections . the activity dependent brain model
@xcite has been implemented on small world networks , by rewiring a small percentage of the square lattice bonds .
again universal scaling behaviour is recovered for both the avalanche distribution and the power spectra .
the simple rewiring procedure , however , only allows long range connections leaving the average node connectivity equal to a few units , as for _ in vitro _ systems . in this paper
we investigate the behaviour of the activity dependent brain model on scale - free networks , whose feature are closer to the morphology of neuronal networks in living brains .
scale - free networks are indeed characterized by a power law distribution of the node connectivity , allowing a high number of connections per neuron .
we develop the model on the apollonian network @xcite , that has the property of being simultaneously small - world and scale - free and therefore exhibits all characteristics found for neuronal networks .
analogously to previous studies , we analyse the behaviour of the avalanche size and duration distribution and the power spectra related to electrical activity .
we also study a system composed by both excitatory and inhibitory sysnapses , to be closer to real brains .
the paper is organized as follows : in sect.[net ] the scale - free apollonian network is described , whereas in section [ mod ] the activity dependent brain model is presented and the results on brain activity are discussed in section [ res ] . concluding remarks are given in section [ conclu ] .
the apollonian network has been recently introduced @xcite in a simple deterministic version starting from the problem of space - filling packing of spheres according to the ancient greek mathematician apollonius of perga . in its classical version
the network associated to the packing gives a triangulation that physically corresponds to the force network of the packing .
one starts with the zero - th order triangle of corners @xmath2 , places a fourth site @xmath3 in the center of the triangle and connects it to the three corners ( @xmath4 ) .
this operation will divide the original triangle in three smaller ones having in common the central site .
the iteration @xmath5 proceeds placing one more site in the center of each small triangle and connecting it to the corners ( fig.1 ) . at each iteration @xmath6 , going from 0 to @xmath7 , then the number of sites increases by a factor 3 and the coordination of each already existing site by a factor 2 .
more precisely , at generation @xmath7 there are @xmath8 vertices with connectivity degree @xmath9 respectively where the two last values correspond to the site @xmath3 and the three corners @xmath2 .
the maximum connectivity value then is the one of the very central site @xmath3 , @xmath10 , whereas the sites inserted at the @xmath7-th iteration will have lowest connectivity 3 . :
iterations @xmath11 are symbols @xmath12 , respectively .
, width=264 ] the important property of the apollonian network is that it is scale - free .
in fact , it has been shown @xcite that the cumulative distribution of connectivity degree @xmath13 , where @xmath14 is the total number of sites at generation @xmath7 , has a power law behaviour with @xmath15 .
more precisely , @xmath16 , with @xmath17 .
moreover the network has small - world features .
this implies @xcite that the average length of the shortest path @xmath18 behaves as in random networks and grows slower than any positive power of @xmath7 , i.e. @xmath19 . furthermore the clustering coefficient @xmath20 is very high as in regular networks ( @xmath21 ) and contrary to random networks . for the apollonian network @xmath20
has been found to be equal to @xmath22 in the limit of large @xmath7 . on this basis
the apollonian network appears to have all the features typical of neuronal networks : small - world property found experimentally @xcite and possibility of very high connectivity degree ( scale - free ) .
moreover it also presents bonds connecting sites of all lengths .
also this last feature is characteristic of neuronal networks in brain cortex , where the length of an axon connecting the pre - synaptic with the post - synaptic neuron can vary over several orders of magnitude , from @xmath23 to @xmath24 .
on a apollonian network at generation @xmath7 , we assign at each site a neuron at potential @xmath25 and at each bond a synapse of conductance @xmath26 .
whenever at time @xmath27 the value of the potential at a site @xmath28 is above a certain threshold @xmath29 , the neuron generates an action potential , distributing charges to connected neurons in proportion to the current flowing through each bond @xmath30 where @xmath31 is the potential at time @xmath27 of site @xmath32 , connected to site @xmath28 , @xmath33 and the sum is extended to all @xmath15 sites connected to site @xmath28 that are at a potential @xmath34 . after firing a neuron
is set to a zero resting potential .
the conductances can be initially all set equal or else random between 0 and 1 , whereas the neuron potentials are uniformly distributed random numbers between @xmath35 and @xmath36 . in agreement with the soc scenario ,
the initial state for the voltage is not relevant since the system evolves toward the same critical state regardless of the initial condition .
the potential is fixed to zero at the three sites 1,2,3 where information can flow out of the system .
the external stimulus can be imposed at one input site chosen either fixed or at random , this last case modelling more closely spontaneous brain electrical activity .
the firing rate of real neurons is limited by the refractory period , i.e. the brief period after the generation of an action potential during which a second action potential is difficult or impossible to elicit .
the practical implication of refractory periods is that the action potential does not propagate back toward the initiation point and therefore is not allowed to reverberate between the cell body and the synapse . in the model , once a neuron fires , it remains quiescent for one time step and is therefore unable to accept charge from firing connected neurons .
this ingredient indeed turns out to be crucial for a controlled functioning of the numerical model . in this way
an avalanche of charges can propagate far from the input through the system .
as soon as a site is at or above threshold @xmath37 at a given time @xmath27 , it fires according to eq .
. then the conductance of all the bonds , connecting to active neurons and that have carried a current , is increased in the following way @xmath38 where @xmath39 , with @xmath40 being a dimensionless parameter and @xmath41 a unit constant bearing the dimension of an inverse potential . after applying eq .
( 2 ) the time variable of the simulation is increased by one unit .
( 2 ) describes the mechanism of increase of synaptic strength , tuned by the parameter @xmath40 .
this parameter then represents the ensemble of all possible physiological factors influencing synaptic plasticity , many of which are not yet fully understood .
once an avalanche of firings comes to an end , the conductance of all the bonds with non - zero conductance is reduced by the average conductance increase per bond , @xmath42 , where @xmath43 is the number of bonds with non - zero conductance .
the quantity @xmath44 depends on @xmath40 and on the response of the brain to a given stimulus . in this way
the network `` memorizes '' the most used paths of discharge by increasing their conductance , whereas the less used synapses atrophy .
once the conductance of a bond is below an assigned small value @xmath45 , it is removed , i.e. is set equal to zero , which corresponds to what is known as pruning .
this remodelling of synapses mimicks the fine tuning of wiring that occurs during `` critical periods '' in the developing brain , when neuronal activity can modify the synaptic circuitry , once the basic patterns of brain wiring are established @xcite .
these mechanisms correspond to a hebbian form of activity dependent plasticity , where the conjunction of activity at the presynaptic and postsynaptic neuron modulates the efficiency of the synapse @xcite . to insure the stable functioning of neural circuits , both strengthening and weakening rules of hebbian synapses are necessary to avoid instabilities due to positive feedback @xcite .
however , differently from the well known long term potentiation ( ltp ) and long term depression ( ltd ) mechanisms , the modulation of synaptic strength does not depend on the frequency of synapse activation @xcite . the external driving mechanism to the system is imposed by setting the potential of the input site to the value @xmath37 , corresponding to one stimulus .
this external stimulus is needed to keep the system functioning and therefore mimicks the living brain activity .
the discharge evolves until no further firing occurs , then the next stimulus is applied .
we consider an apollonian net at the generation @xmath46 ( 29527 neurons and 177150 synapses ) .
the three corner sites of the system have always zero potential and represent open boundary conditions .
the input site is either chosen at random or fixed .
it is worth noticing that the case of random input sites simulates more closely the spontaneous activity of the system .
synapses can be excitatory or inhibitory with probability @xmath47 .
initial conductances can be either all equal to @xmath48 or randomly distributed between 0 and 1 .
the other parameters in the simulation are : firing threshold @xmath49 and conductance cut - off for pruning @xmath50 . their value does not influence the simulation results . _ pruning _ the strength of the parameter @xmath40 , controlling both the increase and decrease of synaptic strength , determines the plasticity dynamics in the network .
in fact , the more the system learns strengthening the used synapses , the more the unused connections will weaken .
we apply a sequence of external stimuli and we measure the total number of pruned bonds at the end of each avalanche , @xmath51 .
this quantity in general could depend on the initial conductance @xmath52 , therefore the two cases of all initial conductances equal to 0.25 , and uniformly distributed between 0 and 1 , are investigated . and equal initial conductances . in the inset , asymptotic number of active bonds as function of @xmath40 .
the maximum is for @xmath53 where are active about @xmath54 of bonds .
, width=302 ] first the case of equal initial conductances is analysed . for each value of @xmath40
the average number of pruned bonds , @xmath51 , is monitored as function of time , where a time unit corresponds to the application of an external stimulus . for input sites
randomly chosen at each stimulus , fig.2 shows that pruning starts after a certain time , since all conductances are initially equal to 0.25 , and @xmath51 increases more rapidly with time for larger @xmath40 .
the plateau is reached after about 5000 stimuli ( for every @xmath40 ) after which @xmath51 increases only of few units in time . from the asymptotic value of each curve we can evaluate the asymptotic number of active bonds as function of @xmath40 and
determine that the value of @xmath40 maximizing the number of active bonds is about 0.020 .
this could be interpreted as an optimal value for the system with respect to plastic adaptation : it maximizes the number of active connections under the competing strengthening and weakening rules . in order to understand
if pruning acts in the same way on bonds created at different iterations @xmath55 , or rather tends to eliminate some particular iteration , the probability to prune bonds of different @xmath6 is evaluated , that is the number of pruned bonds over the total number of bonds for each iteration stage , as function of time .
fig.3 shows that the plateau is reached at about the same time and the shape of the curve is similar for each @xmath6 .
however the probability to prune bonds with large @xmath6 is higher : these are the bonds created in the last iterations and therefore embedded in the interior of the network .
this suggests that the most active bonds are the long range ones ( small @xmath6 ) , that therefore optimize information transport through the network . in the inset of fig.3
we show the asymptotic number of pruned bonds per generation on a semi - log scale , this quantity is well fitted by the exponential behaviour @xmath56 . as function of time for equal initial conductances .
in the inset , the asymptotic @xmath51 ( after 5000 stimuli ) as function of @xmath6 with the exponential fit @xmath57 .
, width=302 ] the same analysis has been performed for random initial conductances between 0 and 1 .
the results are similar to the previous case .
it can be noticed that pruning starts already at @xmath58 , since conductances close to zero are present , and the plateau is reached after about 3000 stimuli .
the value of @xmath40 which now optimizes the number of active bonds is about 0.030 .
finally the pruning behaviour for different iterations is similar to the previous case , with the pruning probability also increasing with @xmath6 exponentially as @xmath56 .
the effect of pruning on the connectivity degree of the network ( fig.4 ) has also been analysed , i.e. the number of sites with a given connectivity degree @xmath15 as function of @xmath15 in the initial network and after application of a given number of external stimuli . in order to identify the different stages in the pruning process
, the inset of fig.4 shows the total number of pruned bonds as function of time . after the application of few external stimuli ,
i.e. for a short plastic training , the distribution @xmath59 shows the same scaling behaviour of the apollonian network .
as the pruning process goes on , sites vary their connectivity and new values of @xmath15 appear .
the result is that the scaling behaviour is progressively lost , as well as the scale - free character of the network , since there is a generalized decrease of connectivity in the network . in the analysis of spontaneous activity it is therefore important to impose a not too extended plastic training in order to avoid an excessive decrease of connectivity degree of the network . at different pruning stages @xmath27 for equal initial conductances and @xmath53 . in the inset the corresponding behaviour of the number of pruned bonds .
, width=302 ] _ spontaneous activity : avalanche distributions _ after training the system applying plasticity rules during @xmath60 external stimuli ,
we now submit the system to a new sequence of stimuli with no modification of synapsis strength .
the response of the system to this second sequence models the brain spontaneous activity , which is analysed by measuring the avalanche size distribution @xmath61 , the time duration distribution @xmath62 , and the power spectrum @xmath63 . , random initial conductances , @xmath64 and random input site . in the inset the corresponding behaviour of the number of pruned bonds . , width=302 ] + 1 cm [ fig.5 ] the avalanche size distribution @xmath61 consistently exhibits power law behaviour for different values of model parameters .
fig.5 shows the avalanche size distribution for different values of @xmath60 , including also the case @xmath65 ( no plasticity training ) for random initial conductances .
we notice that , for fixed size @xmath66 , increasing @xmath60 decreases the number of avalanches of that size , suggesting that strong plasticity remodelling decreases activity .
the exponent appears to be independent of @xmath60 as long as the number of pruned bonds , @xmath51 , is far from the plateau ( see inset in fig.5 ) .
similar results are found for equal initial conductances , the value of the exponent is @xmath67 and is stable with respect to variations of the parameter @xmath40 for both equal and random initial conductance .
this value is compatible within error bars with the value found in the experiments of beggs and plentz @xcite , @xmath68 , and with previous results of the model on both regular and small world lattices .
this suggests that the high level of connectivity reduces the probability of very large avalanches but does not change substantially the spontaneous activity behaviour . for larger
@xmath60 , the distribution exhibits an increase in the scaling exponent and finally looses the scaling behaviour for very large @xmath60 values in the plateau regime for the number of pruned bonds . .
only distributions for small @xmath15 are shown , for higher @xmath15 the scaling behaviour is lost ( random initial conductances , @xmath69 ) .
, width=302 ] it is important to investigate the role of the choice of fixed input site , since in the apollonian network , contrary to the regular network , sites may have very different connectivity degree .
fig.6 shows the avalanche size distribution for input sites randomly chosen among sites with given connectivity degree @xmath15 . in this way it is possible to detect solely the effect due to the connectivity of the input site , eliminating all other effects due to the particular position of the input site in the network .
power law behaviour is found for connectivity degree of the input site up to @xmath70 .
the scaling exponent decreases with increasing connectivity degree @xmath15 of the input site , that is for larger @xmath15 larger avalanches become more probable .
however , if the connectivity degree increases further , the scaling behaviour is lost .
this is due to the fact that an input site with very high connectivity must distribute its charge to many connected sites and therefore the network activity will be damped already at the initial site .
therefore , to reproduce the experimentally observed scaling behaviour , the fixed input site should be chosen with low connectivity degree ( @xmath71 ) .
the avalanche size distribution for fixed input site with connectivity @xmath72 or @xmath73 exhibits power law behaviour with the same exponents found for random input site : @xmath67 for equal and random initial conductances . at time @xmath74 a neuron is activated by an external stimulus initiating the avalanche .
this will continue until no neuron is at or above threshold .
the number of avalanches lasting a time @xmath75 , @xmath62 , as function of @xmath75 exhibits power law behaviour ( fig.7 ) with an exponential cutoff .
the scaling exponent is found to be @xmath76 for equal and random initial conductances .
this value is found to be stable with respect to different @xmath40 ( fig.7 ) and @xmath60 , provided that the number of pruned bonds @xmath51 is lower than the plateau for that value of @xmath40 .
moreover it does not depend on the choice of the input site , either fixed or random .
finally both values agree within error bars with the value 2.0 , exponent found experimentally by beggs and plentz @xcite .
( random initial conductances , random input sites , @xmath77 ) .
the dotted line has slope 2.1.,width=302 ] _ power spectra for spontaneous activity _ in order to compare the results of the apollonian network with eeg medical data , the power spectrum of the resulting time series can be calculated . for this purpose ,
the number of active neurons is monitored as function of time during spontaneous activity .
fig.8 shows an example of neuronal activity where avalanches of all sizes can be generated in response to the external stimulus .
the power spectrum is calculated as the squared amplitude of the fourier transform as function of frequency , averaged over many sequences of external stimuli .
fig.9 shows the spectrum for equal initial conductances and different values of @xmath60 . for @xmath65 ,
i.e. when no plasticity mechanism is applied , the spectrum has a behaviour @xmath78 , characteristic of soc . for values of @xmath60 different from zero but before @xmath51 reaches the plateau , one can distinguish two different regimes : a power law behaviour with exponent @xmath79 at high frequency , followed by a crossover toward white noise at low frequency . however , for @xmath80 ( close to the plateau value for @xmath51 ) the scaling behaviour with exponent 0.8 is detected over a wider frequency range . the difference between @xmath81 for @xmath65 and @xmath82 for higher @xmath60
, suggests that the existence of plasticity rules reduces the power spectrum exponent reaching agreement with experimental eeg spectrum @xcite .
the stability of the exponent with respect to @xmath40 has also been verified , finding consistently @xmath79 at high frequency .
finally the power spectrum for fixed input site shows a scaling exponent @xmath79 over two orders of magnitude . the measured value for the power spectra exponent is in agreement with the expected relation with the scaling exponent of the avalanche duration distribution @xmath83 , being @xmath84 @xcite .
the scaling behaviour of the power spectrum can be interpreted in terms of a stochastic process determined by multiple random inputs @xcite .
in fact , the output signal resulting from different and uncorrelated superimposed processes is characterized by a power spectrum with power law behaviour and a crossover to white noise at low frequencies . the crossover frequency is related to the inverse of the longest characteristic time among the superimposed processes .
the value of the scaling exponent depends on the ratio of the relative effect of a process of given frequency on the output with respect to other processes . @xmath78
noise corresponds to a superposition of processes of different frequency having all the same relative effect on the output signal . in our case
the scaling exponent is smaller than unity , suggesting that processes with high characteristic frequency are more relevant than processes with low frequency in the superposition @xcite . , equal initial conductances , @xmath53 and random input sites . ,
width=302 ] _ inhibitory synapses _ in the mature living brain synapses can be excitatory or inhibitory , namely they set the potential of the post - synaptic membrane to a level closer or farther , respectively , to the firing threshold
. this ingredient can be indroduced by considering each synapse inhibitory with probability @xmath85 and excitatory with probability @xmath86 .
the avalanche size and duration distributions show that the exponents @xmath87 and @xmath88 increase for increasing @xmath85 , therefore for a high percentage of inhibitory synapses the probability of large avalanches decreases ( fig.10 ) . on the regular lattice for @xmath89
no longer power law but exponential behaviour is found @xcite . in the present case scaling behaviour persists due to the very high connectivity degree , suggesting that the apollonian network is better suited to model the neural connections in the brain . the power spectra for different values of @xmath85 exhibit a complex behaviour .
in fact , for a small fraction of inhibitory synapses ( @xmath90 ) the power law exponent @xmath91 increases with respect to the case where synapses are all excitatory up to a value 1.2 .
then , for @xmath92 , the exponent decreases toward values compatible with experimental results , i.e. between 0.7 and 1.0 . by increasing further
the percentage of inhibitory synapses , to values close or greater than 0.2 , the spectrum becomes the one of white noise .
, equal initial conductances , @xmath53 , random input sites and @xmath93 . ,
extensive simulations have been performed for the activity dependent brain model implemented on the scale - free apollonian network .
the results are compared with previous simulations on regular and small world lattices and with experimental data .
we first find the striking result that an optimal value of of the plasticity strength @xmath40 exists with respect to the pruning process .
moreover , it appears that synapses of later generations , deeply embedded in the network , are pruned with higher probability with respect to bonds of the early generations , mostly long range , that optimize information transmission .
moreover the avalanche size distribution shows a power law behaviour with an exponent @xmath67 for equal and random initial conductances .
this value is compatible with @xmath68 , experimentally found for neuronal avalanches and recovered by the model on the square lattice and small world networks .
also the avalanche duration distribution exhibits power law behaviour with an exponential cutoff , in agreement with experimental results of beggs and plentz @xcite .
the exponent has value : @xmath76 for equal and random initial conductance , in agreement with 2.0 found experimentally .
furthermore the power spectrum exhibits power law behaviour at high frequency with @xmath79 , in agreement with experimental data @xcite . at intermediate frequency
the slope becomes greater than unity , crossing over to white noise at low frequencies .
none of the scaling exponents for spontaneous activity in the case of excitatory synapses depends on the particular choice for the length or strength of the plasticity training and are quite stable with respect to the initial conductance configurations .
these results suggest that also on apollonian network universal behaviour found for regular and small world networks @xcite holds .
furthermore , the scale - free apollonian net provides an excellent description both of the morphology and the electrical activity properties of the brain .
|
networks of living neurons exhibit an avalanche mode of activity , experimentally found in organotypic cultures .
moreover , experimental studies of morphology indicates that neurons develop a network of small - world - like connections , with the possibility of very high connectivity degree .
here we study a recent model based on self - organized criticality , which consists in an electrical network with threshold firing and activity - dependent synapse strengths .
we study the model on a scale - free network , the apollonian network , which presents many features of neuronal systems .
the system exhibits a power law distributed avalanche activity .
the analysis of the power spectra of the electrical signal reproduces very robustly the power law behaviour with the exponent 0.8 , experimentally measured in electroencephalograms ( eeg ) spectra .
the exponents are found quite stable with respect to initial configurations and strength of plastic remodelling , indicating that universality holds for a wide class of brain models .
= + 1truecm
|
let @xmath0 be the finite field of characteristic @xmath1 with @xmath2 elements .
an elliptic curve @xmath3 over a finite field @xmath0 is given by the _ weierstrass equation _
@xmath4 where the coefficients @xmath5 are in @xmath0 ; see @xcite for a general background and see @xcite for cryptographic interests on elliptic curves . as usual
, let @xmath6 be the set of @xmath0-rational points on elliptic curve @xmath3 including the point at infinity denoted by @xmath7 .
it is known , see @xcite , that @xmath6 is a finite _ abelian _ group with the neutral element @xmath7 and the cardinality of this group satisfies the _ hasse - weil _ bound as @xmath8 it is also known , see @xcite , that the group structure of @xmath6 is expressed by the group isomorphism @xmath9 where unique integers @xmath10 satisfy @xmath11 for the prime power @xmath12 and positive integers @xmath13 ,
let @xmath14 be the number of distinct elliptic curves @xmath3 over @xmath0 ( up to isomorphism over @xmath0 ) such that @xmath15 .
moreover , let @xmath16 be the the number of distinct group structures of all elliptic curves over the finite field @xmath0 .
in this paper , we give explicit formulas for @xmath17 and @xmath16 , for all prime powers @xmath12 and all possible values of @xmath13 .
we use these formulas to derive tight upper and lower bounds on @xmath16 and aslo an asymptotic formula for the average value of @xmath16 on average over prime powers @xmath18 as @xmath19 .
finally , we also present some numerical results concerning the frequency of the `` most common '' group structure over @xmath0 , that is , for @xmath20 these results reveal several interesting phenomena in the behaviour of this function and also of the parameters @xmath21 , @xmath22 and @xmath23 on which it value is achieved .
finally , we note the distribution of group structures generated by elliptic curves generated by all possibke finite field @xmath0 has been studied in @xcite . throughout the paper
, @xmath1 always denotes a prime and @xmath24 always denotes a prime power .
let @xmath25 be an integer such that @xmath26 and @xmath27 .
let @xmath28 and let @xmath29 be the largest integer such that @xmath30 let @xmath31 be the largest integer such that @xmath32 we note that @xmath33 . for each positive divisor @xmath21 of @xmath31 ,
let @xmath34 and @xmath35 as usual , we use @xmath36 and @xmath37 to denote the number of positive integer divisors and the euler function of @xmath38 , respectively .
moreover , for every negative integer @xmath39 with @xmath40 or @xmath41 we denote by @xmath42 the class number of some quadratic order of discriminant @xmath39 .
the implied constants in the symbols ` @xmath43 ' , ` @xmath44 ' and ` @xmath45 ' are absolute . we recall that the notations @xmath46 , @xmath47 and @xmath48 are all equivalent to the assertion that the inequality @xmath49 holds for some constant @xmath50 .
for @xmath51 , let @xmath52 be the quadratic character modulo @xmath1 .
so , for a positive integer @xmath53 , we have @xmath54 or @xmath55 , if @xmath56 , @xmath57 for some @xmath58 or @xmath59 for all @xmath60 , respectively .
moreover , for @xmath61 , let @xmath62 equals @xmath63 or @xmath55 if @xmath64 , @xmath65 or @xmath66 , respectively .
[ thm : gmn ] let @xmath2 be a power of a prime @xmath1 .
let @xmath21 , @xmath22 be positive integers .
let @xmath67 and @xmath28 .
then , @xmath14 , that is , the number of @xmath0-isomorphism classes of elliptic curves @xmath3 over @xmath0 such that @xmath15 , equals : 1 .
@xmath68 , if @xmath26 , @xmath69 , @xmath70 , and @xmath71 2 .
@xmath72 , if @xmath73 is odd , @xmath74 and @xmath75 3 .
@xmath76 , if @xmath73 is odd , @xmath77 , @xmath78 , and @xmath79 4 .
@xmath80 , if @xmath73 is odd , @xmath61 or @xmath81 , @xmath74 , and @xmath82 5 .
@xmath83 , if @xmath73 is even , @xmath74 , and @xmath75 6 .
@xmath84 , if @xmath73 is even , @xmath74 , and @xmath85 7 .
@xmath86 , if @xmath73 is even , and @xmath87 8 .
@xmath88 , otherwise .
the following result gives explicit formulas for the number @xmath16 of distinct group structures of all elliptic curves over @xmath0 .
[ thm : g ] let @xmath2 be a power of a prime @xmath1 .
for the number @xmath16 of distinct group structures of all elliptic curves over @xmath0 , we have @xmath89 we now present explicit upper and lower bounds on @xmath16 .
[ thm : lug ] let @xmath2 be a power of a prime @xmath1 . for the number @xmath16 of distinct group structures of all elliptic curves over @xmath0 , we have @xmath90 we also show that the bounds of theorem [ thm : lug ] are asymptotically tight .
[ thm : limg ] when @xmath91 via the set of prime powers , we have 1 . @xmath92 ; 2 . @xmath93 ; 3 .
@xmath94 .
finally , we derive an asymptotic formula for the average value of @xmath16 .
[ thm : asympg ] for @xmath95 , when @xmath12 runs over via the set of prime powers , we have @xmath96 where @xmath97 our argument can also be used to obtain an explicit bound on the error term in theorem [ thm : asympg ] .
let @xmath3 be an elliptic curve over @xmath0 of characteristic @xmath1 .
let @xmath98 and @xmath99 .
let @xmath100 denote the _ frobenius _ endomorphism on @xmath3 , that is given by @xmath101 we note that , @xmath100 is the root of the characteristic polynomial @xmath102 in the ring of @xmath0-endomorphisms of @xmath3 ; this ring is denoted by @xmath103 .
moreover , by @xmath104 we denote full endomorphism ring , that is , the ring of @xmath105-endomorphisms of @xmath3 .
let @xmath28 be the discriminant of the characteristic polynomial of @xmath3 .
suppose @xmath26 .
then , @xmath3 is called an ordinary elliptic curve .
we have @xmath106 .
moreover , @xmath107 is isomorphic to some order @xmath43 in the quadratic imaginary field @xmath108 .
in particular , we have @xmath109=\z\[\frac{\delta+\sqrt{\delta}}{2}\ ] \subseteq \end(e ) \subseteq o_k,\ ] ] where @xmath110 is the maximal order in @xmath111 , that is , the ring of algebraic integers of @xmath111 .
let @xmath112 $ ] be the conductor of @xmath113 $ ] , that is the largest integer such that @xmath114 .
then , @xmath115 , called the fundamental discriminant , is the discriminant of the field @xmath111 . also , @xmath116 $ ] .
we note that , @xmath117 , where the conductor @xmath118 $ ] is a divisor of @xmath29 .
furthermore , @xmath119 is the discriminant of @xmath43 , so the order @xmath43 is uniquely determined by its discriminant and denoted by @xmath120 .
we let @xmath121 be the class number of @xmath43 which is also denoted by @xmath122 . now , suppose @xmath123 . then
, @xmath3 is called a supersingular elliptic curve .
let @xmath124 denote the unique quaternion algebra over @xmath125 which is only ramified at @xmath1 and @xmath126 .
then , @xmath103 is either a quadratic order in @xmath108 or a maximal order in @xmath127 .
moreover , @xmath107 is a maximal order in @xmath127 ; see @xcite or @xcite .
two elliptic curves over @xmath0 are called _ isogenous _ over @xmath0 if and only if they have the same number of points over @xmath0 .
the number of @xmath0-rational points of the elliptic curve @xmath3 over @xmath0 satisfies the _ hasse - weil _ bound . on the other hand , _ deuring - waterhouse _
theorem , see @xcite , describes all possible values of @xmath128 that can be the cardinality of @xmath6 , for some elliptic curve @xmath3 over @xmath0 .
let @xmath2 be a power of a prime @xmath1 .
let @xmath129 and let @xmath130 .
the integer @xmath128 is the cardinality of @xmath6 , for some elliptic curve @xmath3 over @xmath0 , if and only if one of the following conditions is satisfied : 1 .
@xmath69 and @xmath26 2 .
@xmath73 is odd and @xmath131 3 .
@xmath73 is odd , @xmath132 , @xmath61 or @xmath81 4 .
@xmath73 is even , @xmath131 , @xmath133 5 .
@xmath73 is even , @xmath134 , @xmath135 6 .
@xmath73 is even , @xmath136 .
see @xcite .
let @xmath137 be a negative integer with @xmath138 or @xmath41 and let @xmath139 be the largest integer such that @xmath140 and @xmath141 or @xmath41 .
let @xmath142 denote the _ kronecker class number _ of @xmath137 .
we have @xmath143 let @xmath144 be the number of distinct elliptic curves @xmath3 over @xmath0 ( up to isomorphism over @xmath0 ) such that @xmath145 .
the following lemma gives explicit formulas for the values of @xmath144 .
[ lem : ison ] let @xmath2 be a power of a prime @xmath1 .
let @xmath129 and let @xmath130 .
then , @xmath144 , that is , the number of @xmath0-isomorphism classes of elliptic curves @xmath3 over @xmath0 with @xmath146 , equals : 1 .
@xmath147 , if @xmath69 and @xmath26 2 .
@xmath148 , if @xmath73 is odd and @xmath131 3 .
@xmath80 , if @xmath73 is odd , @xmath132 , @xmath61 or @xmath81 4 .
@xmath83 , if @xmath73 is even , @xmath131 , @xmath133 5 . @xmath84 , if @xmath73 is even , @xmath134 , @xmath135 6
. @xmath86 , if @xmath73 is even , @xmath136 7 .
@xmath88 , otherwise .
see ( * ? ? ?
* theorem 4.6 ) .
the group of @xmath0-rational points on the elliptic curve @xmath3 over @xmath0 is isomorphic to the group @xmath149 , with unique integers @xmath10 such that @xmath70 and @xmath71 .
we note that , every group @xmath149 , with integers @xmath10 satisfy later conditions , may not occur as the group @xmath6 for some elliptic curve @xmath3 over @xmath0 .
the following theorem describes the possible group structures for elliptic curves over finite fields , see @xcite .
[ lem : gs ] let @xmath2 be a power of a prime @xmath1 .
let @xmath21 , @xmath22 be positive integers with @xmath150 .
let @xmath67 .
there is an elliptic curve @xmath3 over @xmath0 such that @xmath151 if and only if one of the following holds : 1 .
@xmath26 , @xmath27 , @xmath70 and @xmath71 2 .
@xmath73 is odd , @xmath131 , @xmath152 , and @xmath74 3 .
@xmath73 is odd , @xmath131 , @xmath77 , and @xmath74 or @xmath153 4 .
@xmath73 is odd , @xmath132 , @xmath61 or @xmath81 , and @xmath74 5 .
@xmath73 is even , @xmath131 , @xmath133 , and @xmath74 6 .
@xmath73 is even , @xmath134 , @xmath135 , and @xmath74 7 .
@xmath73 is even , @xmath136 , and @xmath154 .
we note that , the case 1 in lemma [ lem : gs ] corresponds to ordinary elliptic curves and the other cases corresponds to suppersingular elliptic curves . as usual
, we let @xmath155 $ ] be the set of @xmath156-torsion points of the elliptic curve @xmath3 over @xmath0 , that is , @xmath157={\left\{p~:~p\in
e(\overline{\f}_q),\ lp=\co\right\}}.\ ] ] we note that , if @xmath158 , then @xmath157 \cong \z / l\z \times \z / l\z.\ ] ] [ lem : endg ] let @xmath3 be an ordinary elliptic curve over @xmath0 .
the following are equivalent : 1 .
@xmath159 \subseteq e(\f_q)\right\}}$ ] 2 .
@xmath160 3 .
@xmath151 , where @xmath70 and @xmath71 .
we recall , ( * ? ? ?
* prposition 3.7 ) , that for all positive integers @xmath156 with @xmath158 , we have @xmath155 \subseteq e(\f_q)$ ] if and only if @xmath161 , @xmath162 and @xmath163 .
therefore , the descriptions of @xmath21 in cases 1 and 2 are the same .
moreover , for all positive integers @xmath156 with @xmath158 , we have @xmath155 \simeq \z_l \times \z_l$ ] .
suppose @xmath151 , where @xmath70 and @xmath71 .
then , for all @xmath156 with @xmath158 , we have @xmath155 \subseteq e(\f_q)$ ] if and only if @xmath164 .
hence , cases 1 and 3 are also equivalent .
we recall the definition of the numbers @xmath29 and @xmath31 and of the sets @xmath165 given in section [ sec : intro ] .
[ lem : ende ] let @xmath3 be an ordinary elliptic curve over @xmath0 .
assume that @xmath10 are positive integers with @xmath166 , @xmath71 and @xmath167 .
then , we have @xmath151 if and only if @xmath168 for some @xmath169 .
we note that @xmath170 where @xmath156 is some positive divisor of @xmath29 . by assumption , @xmath21 is a divisor of @xmath31 .
> from lemma [ lem : endg ] , we have @xmath151 if and only if @xmath21 is the largest divisor of @xmath31 satisfying @xmath171 .
the latter is equivalent to have @xmath172 which completes the proof . for a real @xmath173 and integers
@xmath174 we denote by @xmath175 the number of primes @xmath176 such that @xmath177 .
an asymptotic estimate of the number of primes in arithmetic progressions is given by the _ siegel
walfisz theorem _ ,
see ( * ? ? ?
* theorem 1.4.6 ) .
[ lem : sigwalf ] for every fixed @xmath178 there exists @xmath179 such that for @xmath180 and for all positive integers @xmath181 , @xmath182 where @xmath183
we note that @xmath184 if and only if @xmath13 satisfy one of the cases given by lemma [ lem : gs ] .
so , we study the nonzero number @xmath14 for the possible values of @xmath13 as follows .
for case 1 , we assume that @xmath26 and @xmath69 . from lemma [ lem : gs ] , we see that @xmath184 if and only if @xmath70 and @xmath71 .
so , let @xmath10 be positive integers satisfying the latter conditions . from lemma [ lem : ende ] , for all elliptic curve @xmath3 over @xmath0 , we have @xmath15 if and only if @xmath185 for some @xmath169 .
we also note that , all orders @xmath186 whit @xmath169 , will occur as the endomorphism ring of some elliptic curves over @xmath0 , see ( * ? ? ?
* theorem 4.2 ) .
moreover , the number of @xmath0 isomorphism classes of elliptic curves with @xmath185 is @xmath187 ( e.g. see ( * ? ? ?
* ; * ? ? ?
* theorem 4.5 ) ) .
therefore , we have @xmath188 for case 2 , we have @xmath131 .
moreover , @xmath14 with @xmath74 is the number of cyclic supersingular elliptic curves over @xmath0 with trace @xmath88 ( up to @xmath0-isomorphism ) , that is , @xmath72 , see ( * ? ? ?
* lemma 4.8 ) . for case 3 , we have @xmath131 and @xmath189 .
also , @xmath14 with @xmath78 is the number of non - cyclic supersingular elliptic curves over @xmath0 with trace @xmath88 ( up to @xmath0-isomorphism ) .
this is @xmath190 . for other cases
, we have @xmath191 .
also , all supersingular elliptic curves in the corresponding isogeny class are cyclic .
then , @xmath14 with @xmath74 is the isogeny class number given by lemma [ lem : ison ] .
so , the proof of theorem [ thm : gmn ] is complete .
the possible group structures of elliptic curves over @xmath0 are the groups isomorphic to @xmath192 , for some values @xmath13 described by lemma [ lem : gs ] . for an integer @xmath25 , let @xmath193 be the the number of distinct group structures of elliptic curves over @xmath0 with the trace @xmath25 .
let @xmath25 be a positive integer with @xmath194 . from lemma [ lem : gs ] , we consider the following cases for @xmath25 . 1 .
suppose @xmath26 .
let @xmath130 .
then , the group @xmath192 , for positive integers @xmath13 with @xmath150 , is the group structure of some elliptic curve @xmath3 over @xmath0 with trace @xmath25 if and only if @xmath70 , @xmath71 and @xmath195 .
this is equivalent to have @xmath196 , @xmath71 and @xmath195 .
as before , let @xmath31 be the largest integer such that @xmath197 and @xmath198 .
therefore , there is a one to one correspondence between the group structures of ordinary elliptic curves over @xmath0 with the trace @xmath25 and positive integer divisors of @xmath31 .
so , @xmath199 if @xmath26 .
2 . suppose @xmath200 .
then , we may have @xmath201 or @xmath202 . from lemma [ lem : gs ]
, we see that @xmath203 now , we sum up @xmath204 over all possible integer values of @xmath25 .
we have @xmath205 using and , we obtain the explicit formulas for @xmath16 .
let @xmath206 be the set of integers of the hasse - weil interval , that is , @xmath207 we recall , from the proof of theorem [ thm : g ] , that for every @xmath208 with @xmath209 , there is a bijection between the set of group structures of isogenous elliptic curves @xmath3 over @xmath0 with order @xmath128 and the set of positive divisors @xmath21 of @xmath210 with @xmath196 . for a positive integer divisor @xmath21 of @xmath210 ,
let @xmath211 be the number of distinct group structures @xmath192 of elliptic curves over @xmath0 for some @xmath212 .
in other words , @xmath211 is the cardinality of the set of positive integers @xmath22 where there exists some elliptic curve @xmath3 over @xmath0 with @xmath151 .
clearly , we have @xmath213 here , we express @xmath211 by counting the number of multiples of @xmath214 in @xmath206 . for a positive integer divisor @xmath21 of @xmath210 , let @xmath215 > from the proof of theorem [ thm : g ] and by lemma [ lem : gs ] , for all positive divisors @xmath21 of @xmath210 , we have @xmath216 where @xmath217 next , using and , we obtain @xmath218 we note that @xmath219 + 1 $ ] .
moreover , for all divisors @xmath21 of @xmath210 , if @xmath220 then @xmath221 and if @xmath222 , then @xmath223-\[\frac{4\sqrt{q}}{m^2p}\]-1 \le \#\ch_q(m ) \le \[\frac{4\sqrt{q}}{m^2}\]-\[\frac{4\sqrt{q}}{m^2p}\]+1\ ] ] and so , @xmath224 to obtain an upper bound for @xmath16 , we write @xmath225 one can see that , @xmath226 then , from , we have @xmath227 now , we provide the lower bound for @xmath16 . if @xmath61 , using , we write @xmath228 + 3 > 2\sqrt{q}+2 .\ ] ] for @xmath229 , using , we write @xmath230 we see from that @xmath231 moreover , one can see that @xmath232 if @xmath233 . from , we have @xmath234 furthermore , @xmath235 if @xmath189 .
therefore , @xmath236 which completes the proof . to prove the result of case 1 , let us choose a sufficiently large integer @xmath237 , and let @xmath238 be the least common multiple of all positive integers @xmath239 .
we now choose a prime @xmath240 and put @xmath241 . using and we derive @xmath242 since by for @xmath243 we have @xmath244 we now derive @xmath245 since by the prime number theorem we have @xmath246 taking @xmath247 we obtain @xmath248 for the above sequence of @xmath243 . for case 2 , we recall a result of heath - brown @xcite , which asserts that there are infinitely many primes @xmath1 such that either @xmath249 for a prime @xmath250 or @xmath251 for a primes @xmath252 for some @xmath253 ( one can take @xmath254 , see the proof of ( * ? ? ?
* lemma 1 ) ) .
using and we see that for each such prime @xmath1 and @xmath243 we have @xmath255 finally in case 3 , we recall that if @xmath256 , where @xmath257 is prime then all prime divisors @xmath250 of @xmath210 satisfy @xmath258 ( since @xmath257 is the multiplicative order of @xmath153 modulo @xmath250 , thus @xmath259 ) . in particular for any @xmath71 with @xmath260 we have @xmath261 . hence , as before , and also recalling , for @xmath256 we obtain @xmath262 which concludes the proof .
since there are @xmath263 prime powers @xmath264 with @xmath265 , using the upper bound of theorem [ thm : lug ] we obtain @xmath266 we see from that @xmath267 we recall the well - known estimate on the divisor function @xmath268 see ( * ? ? ? * theorem 317 ) .
thus @xmath269 by lemma [ lem : sigwalf ] , and partial summation , we see that for @xmath270 we have @xmath271 furthermore , for @xmath272 we use the trivial estimate @xmath273 therefore @xmath274 which together with concludes the proof .
here we present some numerical data concerning the values of @xmath275 given by and also about the values of @xmath21 and @xmath22 at which these values are achieved .
furthermore , we concentrate here on prime values @xmath243 .
first of all we note that for any @xmath128 we have @xmath276 where as before @xmath277 is the number of distinct isomorphism classes of elliptic curves @xmath3 over @xmath278 ( up to isomorphism over @xmath278 ) such that @xmath145 ( see lemma [ lem : ison ] ) .
in particular @xmath279 it is well - known that the bounds on the kronecker class number imply that @xmath280 and for all @xmath281 $ ] but maybe at most two of them , we have @xmath282 see , for example , ( * ? ? ?
* proposition 1.9 ) .
thus , recalling , we derive from the inequalities that @xmath283 we see that from that it is natural to study the values of @xmath284 scaled by @xmath285 .
in fact , our experiments with @xmath286 primes @xmath287 show that scaling by @xmath288 is more natural and the ratio @xmath289 stabilises in a reasonably narrow strip between roughly @xmath290 and @xmath291 , see figure [ fig : maxg ] .
we also notice that for all primes checked the value of @xmath284 is always achieved for @xmath293 with @xmath294 ( that is , for curves with cyclic group of points ) .
moreover , for some primes the same value is also achieved for some pairs @xmath293 with @xmath78 . in our experiments
the value of @xmath284 has never been achieved with @xmath295 . for primes @xmath297 , where we have computed @xmath298
this ration has achieved @xmath80 for @xmath299 , where @xmath284 and @xmath300 are achieved with the same value of @xmath25 with @xmath301 .
also , only four times ( for @xmath302 ) the value of @xmath298 has been below @xmath303 .
unfortunately these extreme values of both types are invisible on figure [ fig : gi ] .
we do not know whether these primes are just some sporadic exceptions or whether there are infinitely many such primes .
more generally , it is certainly interesting to evaluate or at least obtain nontrivial theoretic estimates on @xmath304 this may also help to explain the presence of several horizontal lines on figure [ fig : gi ] ( slightly emphasised there to improve their visibility ) .
clearly , one expects that the value of @xmath284 is achieved for @xmath293 for which @xmath305 , where @xmath306 , is small , so that @xmath307 has a large absolute value which leads to a large value of @xmath277 .
however this is offset by the fact that for @xmath128 having many divisors so the value of @xmath277 is `` split '' between @xmath308 values of @xmath309 .
this effect is observed in the numerical results which are presented below , which show that if @xmath310 then @xmath311 is small but not necessary very small . in particular ,
most of the time @xmath284 and @xmath300 are achieved on different values of @xmath25 , namely in about @xmath312 of the cases within the above range of primes @xmath1 ( more precisely for @xmath313 primes out of the total number of @xmath286 primes @xmath287 ) .
furthermore , it seems that the remaining @xmath314 cases in which @xmath284 and @xmath300 are achieved on the same value of @xmath25 , are the ones that mainly ( but not entirely ) responsible for the presence of horizontal lines on figure [ fig : gi ] .
indeed , the same lines are clearly visible on figure [ fig : gi - t ] where the ratios @xmath298 are plotted only if they come from the same value of @xmath25 .
as we have mentioned , we do not have any solid theoretic explanation to the observed facts .
there is certainly more to investigate here , numerically and theoretically , in order to understand the behaviour of @xmath309 .
|
we obtain explicit formulas for the number of non - isomorphic elliptic curves with a given group structure ( considered as an abstract abelian group ) .
moreover , we give explicit formulas for the number of distinct group structures of all elliptic curves over a finite field .
we use these formulas to derive some asymptotic estimates and tight upper and lower bounds for various counting functions related to classification of elliptic curves accordingly to their group structure . finally , we present results of some numerical tests which exhibit several interesting phenomena in the distribution of group structures .
we pose getting an explanation to these as an open problem .
|
-0.5 cm granular gases @xcite are usually described as collections of macroscopic particles with rough surfaces and dissipative interactions . in order to study them , kinetic theories @xcite and numerical simulations @xcite
were applied for special boundary conditions .
the dynamics of the system is assumed to be dominated by two - particle collisions , modeled by their asymptotic states : a collision is characterized by the velocities before and after the contact , and the contact is assumed to be instantaneous . in the simplest model
, one describes inelastic collisions by normal restitution @xmath0 only . however , surface roughness is important @xcite , since it allows for an exchange of translational and rotational energy .
here we briefly sketch a study of a model where a coulomb friction law with coefficient @xmath2 and a tangential restitution coefficient @xmath7 account for tangential inelasticity and friction @xcite .
we first introduce the model , then we describe a mf theory for the driven granular gas of rough spheres , and finally , we compare analytical results and numerical simulations . in the conclusions we discuss possible future refinements of the computational scheme .
-0.5 cm we consider @xmath8 @xmath9-dimensional spheres of mass @xmath10 and diameter @xmath11 interacting via a hard - core potential , confined on a @xmath12-dimensional ( 2d ) layer of linear size @xmath13 , with periodic boundary conditions . inelasticity and roughness are described by a normal restitution @xmath0 , a coulomb friction law with friction @xmath2 , and a tangential restitution @xmath7 which depends on @xmath0 , @xmath2 and the collision angle @xmath14 for sliding contacts and on a maximum tangential restitution @xmath1 for sticking contacts , when the tangential elasticity becomes important .
when two particles @xmath15 and @xmath12 collide , their velocities after collision depend on the velocities before collision through a collision matrix whose elements depend on @xmath0 , @xmath2 , @xmath14 , and @xmath1 .
thus we calculate the momentum change using a model that is consistent with experimental measurements @xcite . from the momentum conservation laws for linear and angular direction , energy conservation , and coulomb s law
we get the change of linear momentum of particle @xmath15 as a function of @xmath0 , @xmath2 , and @xmath7 @xcite .
the change of the normal component of the relative velocity depends on the normal restitution @xmath0 , which is a tunable parameter , while the change of the tangential component of the relative velocity depends on by the tangential restitution coefficient @xmath16 $ ] , where @xmath1 is the coefficient of maximum tangential restitution , @xmath17.the quantity @xmath18 is determined using coulomb s law such that for solid spheres @xmath19 with the collision angle @xmath20 @xcite . here ,
we simplified the tangential contacts in the sense that exclusively coulomb - type interactions , i.e. @xmath21 is limited by @xmath22 , or sticking contacts with the maximum tangential restitution @xmath1 are allowed @xcite .
-0.5 cm we start from the results of huthmann and zippelius @xcite for a freely cooling gas of infinitely rough particles . here
we apply their mf theory to the case of a driven gas of rough particles , which is the most common experimental situation @xcite .
the mf kinetic theory of huthmann and zippelius is formulated for a gas of rough particles with _ constant _ tangential restitution @xmath7 , corresponding to the limit @xmath23 in our model .
it is based on a pseudo liouville operator formalism and on the assumption of a homogeneous state , with a gaussian probability distribution of translational and rotational energies .
the main outcome of this approach is a set of coupled evolution equations for the translational and rotational temperatures @xmath24 and @xmath25 @xcite . here , we write down and solve the mf equations for @xmath24 and @xmath25 for a granular gas of rough particles in which the translational velocities are subjected to a random uncorrelated gaussian driving of variance @xmath26 .
these equations read for a 2d layer of spheres @xmath27 + m \xi_0 ^ 2 \\ \frac{d}{dt } t_{rot}(t ) = \frac{2}{2d-3}\left [ g b t_{tr}^{3/2 } - g c t _ { tr}^{1/2 } t_{rot}\right]\\
g = 4 a n \sqrt{\frac{\pi}{m } } \chi\;,\ ; a = \frac{1-r_n^2}{4}+\frac{\eta}{2}(1-\eta ) \\ b = \frac{\eta^2}{2q}\;,\ ; c = \frac{\eta}{2q } \left ( 1-\frac{\eta}{q } \right ) ~ , \end{array } \label{mfrp}\ ] ] + where @xmath28 , @xmath29 for spheres , @xmath30 is the gas density and @xmath31 is the pair correlation function at contact . + we can use the verlet - levesque @xcite approximation in 2d @xmath32 where @xmath33 is the volume fraction of the gas . for long times
the system approaches a steady state . by imposing @xmath34
we get the equilibrium temperatures @xmath35 + where @xmath36 for @xmath37 and @xmath38 for spherical particles . by linearizing the set of eqs .
( [ mfrp ] ) around @xmath39 and @xmath40 we get the final approach to the steady state : @xmath41 and @xmath42 $ ] .
the quantity @xmath43 is the enskog collision frequency for elastic particles at the temperature @xmath44 , and @xmath45 is a characteristic relaxation time @xcite . the mf eqs .
( [ mfrp ] ) can be applied to the three parameter model in the limit @xmath46 , see fig .
+ however , experimental measurements are well reproduced by this model only for finite @xmath2 . here
we investigate the possibility to describe the effect of finite friction @xmath2 by replacing @xmath7 in the eqs.([mfrp ] ) with its average @xmath47 over the probability density @xmath48 $ ] of the normalized impact parameter @xmath49 , with the condition @xmath50 .
the used simplifying approach ( @xmath51 ) ignores the fluctuations of @xmath7 due to its dependence on the rotational degree of freedom , and we expect that it is correct only in some trivial limiting cases .
nevertheless , this simple approximation allows us to realize that for @xmath52 , corresponding to many experiments , simulations fit well with the modified mf theory , as we will se below .
if the molecular chaos hypothesis is valid @xmath48=1 $ ] , and we get @xmath53 , with @xmath54 .
-0.5 cm here we compare numerical simulations of a randomly driven monolayer of spheres , performed by using an event driven ( ed ) algorithm @xcite , with the mf predictions ( for details see @xcite ) .
every simulation is equilibrated without driving with @xmath55 and @xmath56 . then inelasticity and driving
are switched on .
we used a fixed volume fraction @xmath57 , @xmath58 and different values of @xmath0 , @xmath1 , @xmath2 . in fig .
[ fig1]a - d , the translational and rotational temperatures for fixed @xmath59 , @xmath60 and different values of @xmath1 are rescaled with the mf equilibrium temperatures and plotted versus the rescaled time @xmath61 .
the high value of @xmath2 allows to decouple normal and tangential momentum ( @xmath62 ) .
the agreement is good , even for low positive values of @xmath1 .
for negative @xmath1 , agreement with mf is observed for the translational temperature , while the rotational temperature shows deviations from scaling in the transient phase , although the equilibrium value fits well with mf , for @xmath63 .
this is due to a failure of the approximation @xmath41 for weak coupling between rotational and translational degrees of freedom and finite time . in fig .
[ fig2]a - b we plot the same quantities as in fig .
[ fig1 ] but for @xmath59 , @xmath5 .
we obtain for @xmath1 near to unity significant deviations from mean field theory , while for negative @xmath1 the agreement with mean field theory is very good .
the explanation for this results is that for @xmath64 and high enough @xmath2 , one has @xmath65 , and this correspond to have @xmath66 . in fig .
[ fig3]a - b we show simulation results for @xmath59 and variable @xmath2 for @xmath6 .
these simulations confirm the previous interpretation .
finally , in fig.[fig3]c - d we show simulation results for @xmath5 , @xmath6 and different @xmath0 . in this case
the data collapse is very good although deviations from mf are observed .
this is a coincidence , since @xmath67 . to improve the mf theory for the finite @xmath2 case
, the effect of a random @xmath7 must be fully taken into account .
we actually are studying the possibility to include a collision angle dependent @xmath7 in the ensemble average of the energy variation due to collisions @xcite .
the mf theory we obtain seems to be very promising and gives a good qualitative agreement with simulations .
a paper is in preparation with the results of this study @xcite and also a three - dimensional analysis is in progress @xcite
. + 99 -0.7 cm h. j. herrmann , j .-
hovi , and s. luding , eds .
, _ physics of dry granular media _ , kluwer academic publishers , dordrecht , netherlands , 1998 .
j. t. jenkins and m. w. richman , phys . of fluids * 28 * , 3485 ( 1985 ) .
m. huthmann and a. zippelius , phys .
e * 56 * , r6275 ( 1997 ) ; s. luding , m. huthmann , s. mcnamara and a. zippelius , phys . rev .
e 58 , 3416 - 3425 , ( 1998 ) .
t. p. c. van noije et al .
* 79 * , 411 ( 1997 ) ; t. p. c. van noije and m. h. ernst granular matter * 1*/2 , 57 - 64 ( 1998 ) ; t. p. c. van noije et al .
cond - mat/9810251 .
s. luding , phys .
e * 52 * , 4442 ( 1995 ) ; s. luding , collisions & contacts between two particles , in ref .
@xcite ; s. luding , e. clment , j. rajchenbach , and j. duran , europhys . lett . * 36 * , 247 ( 1996 ) .
s. f. foerster and m. y. louge and h. chang and k. allia , phys .
fluids * 6 * , 1108 ( 1994 ) ; o. r. walton in _ particulate two - phase flow _
m. c. roco , butterworth , london 1993 , p. 884 .
a. kudrolli , m. wolpert and j. p. gollub , phys .
lett . * 78 * , 1383 ( 1996 ) ; j. s. olafsen and j. s. urbach , phys . rev
. lett . * 81 * , 4369 ( 1998 ) ; w. losert et al .
, cond - mat/9901203 . l. verlet , d. levesque , mol . phys . * 46 * , 969 ( 1982 ) .
r. mazighi , b. bernu , and f. delyon , phys .
e * 50 * , 4551 ( 1994 ) ; s. mcnamara and w. r. young , phys .
e * 50 * , r28 ( 1994 ) .
r. cafiero , s. luding and h. j. herrmann , in preparation .
o. herbst , m. huthmann , and a. zippelius , preprint , submitted to granular matter .
|
we propose a mean field ( mf ) theory for a homogeneously driven granular gas of inelastic particles with coulomb friction .
the model contains three parameters , a normal restitution coefficient @xmath0 , a maximum tangential restitution coefficient @xmath1 , and a coulomb friction coefficient @xmath2 .
the parameters can be tuned to explore a wide range of physical situations . in particular , the model contains the frequently used @xmath3 limit as a special case . the mf theory is compared with the numerical simulations of a randomly driven monolayer of spheres for a wide range of parameter values . if the system is far away from the clustering instability ( @xmath4 ) , we obtain a good agreement between mean field and simulations for @xmath5 and @xmath6 , but for much smaller values of @xmath0 the agreement is less good .
we discuss the reasons of this discrepancy and possible refinements of our computational scheme .
kinetic and transport theory of gases , computational methods in fluid dynamics pacs : 47.50+d , 51.10.+y , 47.11.+j
|
quasar absorption line systems ( qals ) are unique and powerful tools for studying the chemical content , kinematics , ionization state , and overall structure of galaxies as well as of the intergalactic medium over @xmath14 .
absorption is an important feature in many qals , because it relates to metal forming processes .
the mg0.1emii @xmath15 doublet is a strong transition and is easily detected in low ionization qals , making it an extremely useful tool for probing galaxies and their environments and for studying the structure of the intergalactic medium .
weak absorbers are defined to be those absorbers with @xmath0 , and at least some of them represent a unique population(s ) of absorbers that appears to be much different from stronger absorbers @xcite . unlike strong absorbers , which are typically associated with luminous galaxies ( within @xmath16 kpc ) @xcite ,
weak absorbers at @xmath9 are not typically found within @xmath17 kpc of such a galaxy ( * ? ? ?
* but see @xcite for some exceptions ) , raising questions about their processes of formation . through both
statistical means @xcite and through investigation of individual systems @xcite , weak absorbers are found to arise in sub - lyman limit systems ( @xmath18 $ ] ) , comprising at least 25% of forest clouds in that column density range @xcite .
weak absorbers are also found to have high metallicity ; at least 10% solar , but in some cases even solar or supersolar @xcite .
this is a striking result because , as noted above , luminous galaxies are rarely found within a @xmath17 kpc impact parameter of the quasar .
furthermore , some absorbers have / ratios that do not allow for @xmath19-enhancement solely from type ii supernovae , suggesting that metals are also produced `` _ _ in situ _ _ '' by type ia supernovae and not ejected by nearby , luminous galaxies .
it is possible that some weak absorbers arise in areas of intergalactic star formation , or are related to supernovae remnants in dwarf galaxies @xcite .
their incidence is therefore likely to be related to global star formation rates and supernovae activity in these environments .
@xcite have found that a flattened geometry for weak absorbers is most consistent with their number statistics and kinematics , and suggest that such structures could be part of the cosmic web .
photoionization modeling suggests that the absorption arises in a region @xmath20 pcs thick @xcite .
this thin region that produces absorption is most likely surrounded by a lower density region that produces high ionization absorption , with both regions centered at the same velocity @xcite .
components offset in velocity by tens of @xmath21 also often exist in and could represent additional surrounding low density filaments @xcite .
previous surveys for weak absorbers have been conducted at @xmath7 by @xcite and at @xmath9 by @xcite .
@xcite used e140 m space telescope imaging spectrograph _ hubble space telescope _
archive data to search for @xmath22 and @xmath23 as tracers of , and found a number density of absorbers @xmath6 .
@xcite used the high resolution spectrograph on the keck i telescope and observed the mg0.1emii @xmath15 doublet directly , finding @xmath8 .
these values imply that the total cross section for weak absorption is twice that of strong absorbers @xcite .
the findings of @xcite for @xmath7 weak absorbers are consistent with cosmological evolution of the @xmath10 weak absorber population , based upon @xmath24cdm cosmology @xcite .
however , when the effect of the evolution of the extragalactic background radiation ( ebr ) on the absorber population is taken into account , @xmath25 is expected to be higher at @xmath26 than observed .
this is partly because the less intense ionizing background ( compared to @xmath10 ) leads to an increase in the low - ionization phase , i.e. absorbers that were below the equivalent width threshold at @xmath10 are detectable at @xmath26 .
also , there is a significant contribution at @xmath26 from absorption in a lower density phase that gave rise only to higher ionization absorbers at @xmath10 @xcite .
it appears that the processes that lead to weak absorber formation are less active at @xmath26 then at @xmath10 .
determining the evolution of @xmath25 at higher redshifts is an important step toward identifying the formation processes of weak absorbers , and thus served as the motivation for this work . in order to push observational measurements of @xmath25 for weak absorbers to higher redshifts
, we surveyed the spectra of 18 qsos obtained with the ultra - violet echelle spectrograph ( uves ) on the very large telescope ( vlt ) .
such a measurement will shed light on the nature of the absorber population , and may indicate which factors affect the evolution of the absorbers .
the data and survey method are outlined in [ sec : survey ] .
our survey results are summarized in [ sec : results ] .
we also simulated the evolution of the @xmath11 absorber population back to @xmath12 , subject to the changing ebr in order to predict the expected @xmath25 evolution .
these results are presented in [ sec : evol ] and compared with our survey results . in [
sec : conc ] we summarize our findings and end with a discussion .
the 18 uves / vlt qso spectra used for our survey of weak absorption were obtained from the eso archive .
the spectra typically covered a wavelength range of @xmath27 , though there were usually gaps in coverage from @xmath28 and @xmath29 ( see figure [ fig : redcovplot ] ) .
the resolution was @xmath30 , or @xmath31 .
data reduction procedures are described in @xcite .
continuum fitting followed standard procedures using the irafsfit task .
a cubic spline function was used , typically of order 7 - 9 . though the spectra were of extremely high quality
, there were some complicating issues .
as already mentioned , there were some small breaks in the coverage in many of the spectra .
the dense forest made a @xmath32 mg0.1emii @xmath15 doublet survey impractical for some quasars , but this was not a concern since the @xcite survey also covered this range .
there was also significant contamination from telluric absorption features , making a survey of affected wavelengths impractical , as well .
our general philosophy was to exclude troublesome regions rather than to risk missing doublets in our survey . for our survey , we first assumed that every line detected above a @xmath33 significance level was @xmath34 .
it was then determined if the corresponding @xmath35 line was present at the expected location with at least a @xmath36 significance level .
if this was the case , the line profiles were checked for similarity in shape by visual inspection , and for consistency with a 2:1 optical depth ratio . in all cases
some supporting lines at the redshift of our candidate were found ( such as , , , and/or ) .
once the detection of a absorber was confirmed , we measured the rest frame equivalent width . in order to be accepted as a weak absorber , a rest - frame equivalent width of @xmath5 was required .
we note that our survey could have extended to a much lower equivalent width limit .
the @xmath3 limit was chosen for comparison to previous surveys .
this corresponds to @xmath37 $ ] for a typical doppler parameter of @xmath38 , for an unsaturated line .
fortunately , above this limit interference by telluric absorption did not affect our survey in our selected wavelength regions ( i.e. telluric features were weaker than @xmath3 ) . for the purpose of the present survey , we focused on absorbers whose redshift fell in the range @xmath1 .
several quasars had coverage at @xmath39 , but the path length was not significant at that @xmath40 .
the large wavelength coverage of these spectra also allowed us to search for weak absorbers to significantly lower redshifts in most spectra .
therefore , we also considered absorbers with redshifts in the range @xmath9 for comparison with the results of @xcite .
the excellent s / n of these spectra allowed our survey to be nearly 100% complete if we eliminated wavelength ranges with heavy atmospheric absorption .
the telluric features in the eliminated regions acted as `` effective noise '' which made unambiguous detection of a @xmath5 @xmath34 line at a @xmath41 significance level difficult , if not impossible .
figure [ fig : redcovplot ] shows the redshift coverage for the mg0.1emii @xmath15 doublet in each of our spectra . because we excluded certain redshifts from our search and because not every qso spectrum covered the mg0.1emii @xmath15 doublet up to @xmath42 , our average redshift does not coincide with the center of our redshift range .
considering these factors , the average redshift is given by @xmath43 in order to verify that we have nearly 100% completeness in the spectral regions that we surveyed , we randomly added two artificial doublets to each of our spectra , in the redshift range @xmath44 .
simulated lines were not placed in regions of the spectra that were not searched , i.e. the forest , telluric absorption regions , or coverage gaps .
the artificial lines were all assigned the minimum equivalent width for our survey , @xmath3 .
lines were added , beginning with the observed flux and convolving it with the line spread function for the simulated line .
this allowed us to retain a realistic representation of noise in the region , making artificial lines indistinguishable from real ones .
the spectra were searched for these artificial lines in the same manner that we searched for real lines .
all the artificial lines were found even though they were at the minimum equivalent width for our survey .
therefore , we are confident that we effectively have 100% completeness in the regions we searched .
survey completeness as a function of redshift was calculated according to the formalism given in @xcite and @xcite .
the total redshift path covered in the 18 qso spectra over the range @xmath1 is given by @xmath45 where @xmath46 is the function that gives the number of lines of sight along which the mg0.1emii @xmath15 doublet could have been detected at a redshift _ z _ and with a rest frame equivalent width @xmath47 at greater than or equal to a @xmath33 significance level .
_ r _ is the expected ratio of equivalent widths between @xmath34 and @xmath35 , which , according to atomic physics , is two for unsaturated lines . as described in
[ sec : method ] , we find 100% completeness for all doublets with @xmath48 . therefore , we have the simplest case in which @xmath49 , the full path length covered by our survey . for the @xmath1
doublets found in our survey , table [ table : sysprop ] lists the absorber redshifts ( calculated from the optical depth mean ) and the quasar in whose spectrum they were detected .
column 3 lists the velocity width of the systems ( as defined in @xcite ) , column 4 the rest frame equivalent width of @xmath34 , and column 5 the doublet ratio , @xmath50 .
their @xmath34 and @xmath35 profiles are shown in figures [ fig : he2347a]-[fig : he0940 ] .
we find 9 systems over a redshift path of @xmath2 at @xmath51 in the 18 spectra surveyed .
because our survey is 100% complete to @xmath52 , we can simply divide the number of systems by the redshift path to obtain a redshift number density @xmath4 .
figure [ fig : dndzplot ] shows our result , along with the results of @xcite and @xcite .
we also show our survey results for the redshift range @xmath9 , separated into the same bins as in @xcite , for which we find agreement to within @xmath53 .
overall , for @xmath9 , we found 29 systems in a redshift path of @xmath54 , yielding @xmath55 .
@xcite found @xmath8 for a somewhat larger redshift path of @xmath56 .
the solid curve shows the no - evolution expectation for @xmath25 in a @xmath24cdm universe ( @xmath57 and @xmath58 ) normalized to @xmath59 at @xmath60 .
the number density of weak absorbers changes with redshift for a variety of reasons .
since we expect weak absorption to arise in small - scale regions , the process of hierarchical structure growth could destroy the absorbers through mergers or collapse . as the metallicity of the universe increases ,
we expect to see an increase in the number of metal - rich systems
. also , the cumulative effects of supernovae , causing superbubbles and superwinds , will redistribute gas in the universe .
the ionization state of weak absorbers is also expected to change as the ebr evolves .
the ebr is known to change with redshift because of the change in the number density of bright quasars and in the rate of global star formation @xcite .
this changing ebr will have an effect , even on an otherwise static population of absorbers , because of their changing ionization state .
we explore this effect in the following sections .
models by @xcite and @xcite suggest that weak absorbers consist of a high density region of and a lower density phase .
typical densities are on the order of 0.1 @xmath61 for the region that gives rise to the majority of the absorption and 0.001 @xmath61 for the region .
we conduct a thought experiment to explore how these absorbers evolve .
let us assume that a given absorber is stable over time , i.e. there is no gravitational collapse , fragmentation , mergers , evaporation , or pressure changes .
in other words , assume that the absorber is subject to no changes except for the evolving ebr .
the total hydrogen column density , @xmath62 , the electron number density , @xmath63 , and the metallicity , @xmath64 , are all constant .
one may ask how the ionization state of the absorber will change as a result of the changing ebr , between @xmath11 and @xmath12 . at @xmath12
the ebr is @xmath65 dex more intense than at @xmath11 , though the shape of the spectrum is similar .
this increase in the photon number density , @xmath66 leads to a change in the ionization parameter , @xmath67 . for a given cloud we can fix @xmath62 and calculate the change in @xmath68 and @xmath69 . at higher redshift
, we find that the increased ebr causes the equivalent width to be larger for the high ionization state and smaller for the low ionization state , compared to intermediate redshifts . to illustrate the evolution of weak absorbers , we simulate the appearance at @xmath12 of three systems modeled by @xcite at @xmath70 subject to the changing ebr and cosmological evolution .
we applied photoionization models using the code cloudy ( version 94.00 ; @xcite ) with an ebr of the form prescribed by @xcite .
we adopted their model including quasars and star forming galaxies , with an escape fraction of 10% and including absorption by the intergalactic medium .
the results are presented in figures [ fig : sim1 ] - [ fig : sim3 ] .
each of the three systems has only a single cloud detected in , but their absorption profiles differ in strength , kinematic spread , and number of clouds .
the @xmath71 ( figure [ fig : sim2 ] ) and @xmath72 ( figure [ fig : sim3 ] ) systems have offset components in addition to those centered on the . each system was evolved to @xmath12 subject to the changing ebr .
once the column density of each component was found , we created a synthetic , noise - free spectrum . for the @xmath72 system
, the profile evolves below the equivalent width cutoff for our survey ( see figure [ fig : sim3 ] ) .
the overall trend for all three systems is for weak absorbers to be less prevalent at @xmath12 than at @xmath73 , while absorption , which traces the high ionization state , becomes stronger .
we apply a simple physical model to predict @xmath25 at @xmath12 using the results of @xcite as a referemce point .
this model only takes the evolving ebr into account but is useful for exploring which factors affect the evolution of @xmath25 for weak absorbers .
based on the previous measurement of @xmath25 at @xmath11 by @xcite , we now consider what @xmath25 would be expected at @xmath12 subject to the changing ebr .
we first determine the @xmath47 of a @xmath11 absorber whose ionization state has increased just enough from @xmath11 to @xmath12 that it falls below the detection threshold of our survey .
we compute the column density of for an absorber with the threshold @xmath3 to be @xmath74 $ ] , assuming a typical doppler parameter , @xmath38 .
models of weak absorbers @xcite have yielded a range of metallicities , @xmath75 , and a range of densities @xmath76 @xmath61 . to illustrate our method we first choose the values that yield the
smallest predicted @xmath25 value at @xmath77 , @xmath78 and @xmath79 , and solve for the total column density @xmath80 $ ] subject to the haardt - madau ebr at @xmath12 , again using cloudy .
we then compute the @xmath47 of an absorber with this same total column density subject to the ebr at @xmath11 .
we find @xmath81 .
this is the lowest equivalent width absorber at @xmath11 that would have been observed as a weak absorber at @xmath12 .
a similar procedure was followed to find an upper limit on @xmath47 for absorbers that exceed the 0.3 threshold for weak absorption at @xmath11 but were weak at @xmath12 .
such an absorber will have a total column density @xmath82 $ ] , which will produce a @xmath83 absorption at @xmath11 .
once an upper and lower limit on equivalent width at @xmath11 have been obtained , we may integrate over the equivalent width distribution function at this redshift to estimate the number of absorbers per unit redshift that we would expect to observe at @xmath12 .
the equivalent width distribution follows a power law form : @xmath84 where @xmath85 and @xmath86 @xcite .
the expected ratio of @xmath25 at @xmath12 to @xmath25 at @xmath11 , due only to the changing ebr , is given by @xmath87 cosmological evolution of the absorber population must also be taken into account .
for a model with @xmath58 and @xmath57 , cosmological evolution alone would predict a factor of 1.33 more absorbers at @xmath12 .
thus , when we combine evolution due to the ebr and cosmology , we find a predicted @xmath88 at @xmath12 .
if we vary the input cloud properties , such that @xmath89 and @xmath90 , we find a predicted @xmath91 at @xmath77 . in that case
, the electron density change dominates , and the metallicity change only has a small effect . for this estimate we assumed that typical weak doublets
are resolved so that @xmath92 is a typical value
. there may be some unresolved saturation in some of the lines in our sample , since the doublet ratios of weak lines are not equal to @xmath93 ( see table [ table : sysprop ] ) .
assuming @xmath94 , we see no significant change in our predicted @xmath25 .
also , @xmath94 implies @xmath95 k , somewhat smaller than the equilibrium temperature cloudy finds for our assumed density and metallicity .
an even smaller @xmath96 , which we consider unlikely , would result in somewhat smaller @xmath25 predictions , but would also require a very cold cloud .
thus we believe that @xmath92 provides a realistic estimate .
the predicted @xmath25 of @xmath97 is significantly higher than the observed value of @xmath98 for weak absorbers at @xmath99 .
this implies that the population of objects that give rise to weak absorption are becoming more common at @xmath100 than they were at @xmath77 .
this could occur either through gradual formation of these objects , assuming that they are not destroyed , or through an increase in the importance of processes that generate a non - static population .
we have conducted a survey of 18 qso spectra in the eso archive obtained with the uves on the vlt , searching for weak absorbers with @xmath5 . in order to test our survey completeness
we randomly added simulated mg0.1emii @xmath15 transitions to the spectra .
these simulated lines were at the low end of the equivalent width limit . because the spectra have high signal - to - noise , the survey is nearly 100% complete in the regions that we searched , having eliminated regions strongly affected by telluric absorption .
we found 9 systems in a redshift path length of @xmath2 over the range @xmath1 .
this yields a number density per unit redshift of @xmath4 at @xmath99 . for comparison ,
@xcite find @xmath6 at @xmath101 and @xcite find @xmath102 at @xmath70 .
we conclude that the weak absorber population peaks at @xmath10 , and that the objects that generate this type of absorption are less common both at higher and at lower redshifts . a decrease in @xmath25 from @xmath77 to @xmath100
would be expected simply on the basis of cosmological evolution .
the strength of absorption coming from a population of objects is also expected to change subject to the changing ebr . in order to understand the balance of these effects , we calculated the expected @xmath25 at @xmath12 based upon the known population at @xmath11 . for the purpose of this estimate
, we assumed that the absorber population was static or that destruction was balanced by the formation processes between these two epochs , i.e. we only accounted for evolution due to the changing ebr and due to cosmological evolution .
our prediction of @xmath103 is markedly higher than our observed value .
therefore , we conclude that one of two general scenarios must apply . first , the structures that give rise to weak absorption could be long - lived , and there could be a gradual build - up of these structures over time , such as through hierarchical structure formation .
second , the structures could be transient , such that the rate of their formation process determines @xmath25 .
an obvious example of such a process is star formation .
this second possibility was advocated by @xcite who considered the evolution of @xmath25 from @xmath100 to @xmath104 . in that case , starting with the observed @xmath100 population of weak absorbers , the evolving ebr along with cosmological evolution would predict a significantly larger population at @xmath104 than is observed .
this issue was complicated by the incidence of a new type of low redshift weak absorber descended from absorbers from which only absorption was observed at @xmath100 .
nonetheless , the trend for an overall decrease in the weak absorber population is consistent with the decrease in the global star formation rate over that redshift interval .
this suggests a possible relationship between star formation and weak absorbers , which is also consistent with their high metallicities and ( in at least some cases ) iron abundances @xcite .
we should therefore consider how the global star formation rate relates to the results of our @xmath99 survey .
the global star formation rate is roughly constant over the redshift range @xmath105 , and then decreases to lower redshifts ( see @xcite and references therein ) .
if production of weak absorbers was directly correlated with the global star formation rate we would expect a @xmath25 at @xmath77 consistent with a constant creation rate , subject to the ebr and to cosmological evolution .
our observation of a smaller @xmath25 than this expectation would suggest that the global star formation rate should be smaller at @xmath77 than at @xmath100 .
this is not the case .
however , we need to consider that weak absorbers are preferentially found in certain environments .
theoretically , in the context of hierarchical structure formation and empirically on the basis of the morphology density relationship , it is expected that star formation peaks later in low mass galaxies @xcite .
more directly , @xcite found a shift over the range @xmath106 in the relative contributions of low and high mass galaxies to the global star formation rate , with low mass galaxies contibuting more at low redshift .
although measurements of the dwarf galaxy contribution have not yet extended to higher redshifts , it seems likely that the star formation rate in dwarfs peaks at a lower redshift than that for giants .
given the constant global star formation rate at @xmath107 @xcite , if the giants are contributing relatively more at high redshift , the dwarfs must contribute less .
given previous suggestions that weak absorbers are related to dwarfs , it is intriguing that the peak epoch of weak absorbers may coincide with the peak of star formation in the dwarf environment .
@xcite estimate that weak absorbers account for approximately 25%-100% of forest clouds with @xmath108^{-2}$ ] at @xmath13 . at first glance
it seems surprising that the redshift path density for forest clouds would sharply decrease with decreasing @xmath40 until @xmath109 ( @xmath110 ) @xcite , while we find the weak absorber population to increase over the same interval .
however , if the weak absorbers are closely related to star formation in dwarfs , as we have suggested , then there is no reason to expect a direct correlation of their evolution with the evolution of the forest , the latter which is explained by cosmological expansion .
this work was funded by the national science foundation grant nsf ast-07138 and by an nsf reu supplement .
we are indebted to the eso archive for making this work possible .
the authors wish to express their gratitude to chris churchill for his valuable insight and discussions during this project .
thanks also to mike eracleous for his assistance with our analysis .
we also acknowledge joe masiero for providing us with tools for our data processing and andrew mshar for technical assistance .
we also wish to thank an anonymous referee for invaluable and helpful suggestions .
haardt , f. , & madau , p. 2001 , in clusters of galaxies and the high redshift universe observed in x - rays , ed .
d. m. neumann & j. t. t. van ( paris : recontres de moriond xxxvi ) , in press ( astro - ph/0106018 ) steidel , c. c. 1995 , qso absorption lines , proceedings of the eso workshop held at garching , germany , 21 - 24 november 1994 , edited by georges meylan .
springer - verlag berlin heidelberg new york .
also eso astrophysics symposia , 1995 . ,
p.139 , 139 lcrcc he2347 - 4342 & 1.405362 & @xmath111 & @xmath112 & @xmath113 + q0002 - 422 & 1.446496 & @xmath114 & @xmath115 & @xmath116 + q0122 - 380 & 1.450109 & @xmath117 & @xmath118 & @xmath119 + he2217 - 2818 & 1.555845 & @xmath120 & @xmath121 & @xmath122 + he0001 - 2340 & 1.651462 & @xmath123 & @xmath124 & @xmath125 + he0151 - 4326 & 1.708494 & @xmath123 & @xmath126 & @xmath127 + he2347 - 4342 & 1.796237 & @xmath128 & @xmath129 & @xmath130 + q0453 - 423 & 1.858380 & @xmath131 & @xmath132 & @xmath133 + he0940 - 1050 & 2.174546 & @xmath134 & @xmath135 & @xmath136 +
|
the exact nature of weak absorbers ( those with @xmath0 ) is a matter of debate , but most are likely related to areas of local star formation or supernovae activity outside of giant galaxies . using 18 qso spectra obtained with the ultra - violet echelle spectrograph ( uves ) on the very large telescope ( vlt ) ,
we have conducted a survey for weak absorbers at @xmath1 .
we searched a redshift path length of @xmath2 , eliminating regions badly contaminated by atmospheric absorption so that the survey is close to 100% complete to @xmath3 .
we found a total of 9 weak absorbers , yielding a number density of absorbers of @xmath4 for @xmath5 .
@xcite found @xmath6 at @xmath7 and @xcite found @xmath8 at @xmath9 .
therefore , the population of weak absorbers appears to peak at @xmath10 .
we explore the expected evolution of the absorber population subject to a changing extragalactic background radiation ( ebr ) from @xmath11 to @xmath12 ( the median redshift of our survey ) , and find that the result is higher than the observed value .
we point out that the peak epoch for weak absorption at @xmath13 may coincide with the peak epoch of global star formation in the dwarf galaxy environment .
|
the variations of wind speed , in a certain site , are strictly related to the economic aspects of a wind farm , such as maintenance operations , especially in the off shore farms , pitch angle control on new wind turbines and evaluation of a new site .
many scholars have proposed new models that can allow the prediction of wind speed , minutes , hours or days ahead .
many of these models are based on neural networks @xcite , autoregressive models @xcite , markov chains @xcite , hybrid models where the previous mentioned models are combined @xcite , @xcite and other models @xcite .
often , these models are either focused on specific time scale forecasting , or synthetic time series generation .
the approach we propose here is based on indexed semi - markov chain ( ismc ) model that was advanced by the same authors in @xcite and applied to the generation of synthetic wind speed time series . in @xcite
we showed that our model is able to reproduce correctly the statistical behavior of wind speed .
the ismc model is a nonparametric model because it does not require any assumption on the form of the distribution function of wind speed . in this work
we use the same model to forecast future values of wind speed .
we will show that this model performs better than a simple persistence model , by comparing the root mean square errors and the mean absolute errors .
the ismc model is able to forecast wind speed at different time scale without loosing the goodness of forecasting which is almost independent from the time horizon .
the economic aspect of the wind turbine energy production is taken into account by evaluating financial indicators of an investment in wind farm ( internal rate of return , duration and convexity ) .
we also consider two different cases : the first one by using the energy price adding the government incentives and the second without them .
the paper is organised as follows : first , in section 2 , we describe the database and the used commercial wind turbine .
next , in section 3 , basic notation and definitions on the indexed semi - markov chain model are provided .
section 4 is the most important part of the paper ; it contains the relevant results about wind speed forecasting .
continuing , section 5 demonstrates the economic application of the model to a real dataset .
finally , section 6 presents some concluding remarks .
the database used in this work is the same used in our precedent work @xcite , in which it is possible to find a more accurate description and additional information . in this analysis , for the validation of the forecasting model and in order to analyze the behavior at different time scales , we need to resample the original data at different sampling frequencies : namely 30 minutes , 1 hour and 2 hours . in this work
we are going to use the same database also to test our model in the production of energy from a commercial wind turbine which has the blades at a given altitude .
then , we need to consider that the altitude from the ground influences the speed of wind . in particular , it is well known in the literature ( see @xcite ) that the wind speed has the following dependence from the altitude : @xmath0 where @xmath1 is the wind speed at the height of the wind turbine hub , @xmath2 is the value of the wind speed at the height of the instrument , @xmath3 and @xmath4 are respectively the height of the wind turbine and of the instrument ( @xmath5 and @xmath6 ) and @xmath7 is a factor that takes into account the morphology of the area near the wind turbine .
this parameter for a region in which there are no buildings or trees varies from 0.01 to 0.001 , instead for the offshore application it is equal to 0.0001 . for our analysis
we consider a mean value for an onshore application , then we choose @xmath8 . in figure [ fig1 ]
we show the real database and its probability density function that follows a weibull distribution with the following form : @xmath9 where @xmath10 is the shape parameter and @xmath11 is the scale parameter .
the two parameters have been estimated by a maximum likelihood procedure and , for our database are @xmath12 and @xmath13 .
we also estimated the two parameters for the transposed database , we found that the shape parameter remains the same while the scale parameter is @xmath14 .
the semi - markov model we use in our analysis is based on a discrete state space which means that uses only discretized wind speed .
for this reason we select 8 discretized wind speed states ( see table [ st ] ) chosen to cover all the wind speed distribution .
table [ st ] shows the wind speed states with their related wind speed ranges .
1 & 0 to 3 + 2 & 3 - 4 + 3 & 4 - 5 + 4 & 5 - 6 + 5 & 6 - 7 + 6 & 7 - 8 + 7 & 8 - 9 + 8 & @xmath159 + the transformation between wind speed and produced energy is made through a commercial 10 kw aircon hawt wind turbine .
the power curve of this turbine is plotted on figure [ pc1 ] .
the power curve of a specific wind turbine represents how it produces energy as a function of the wind speed values .
the general formulation of the ismc model has been developed in @xcite here we only discussed it informally .
semi - markov processes have similar idea as those that generate markov processes .
the processes are both described by a set of finite states @xmath16 whose transitions are ruled by a transition probability matrix .
the semi - markov process differs from the markov process because the transition times @xmath17 are generated according to random variables .
indeed , the time between transitions @xmath18 is random and may be modeled by means of any type of distribution functions . in studies concerning wind speed modeling the states @xmath16 indicates discretized wind speed at the nth transition and @xmath17 the time in which the nth change of wind speed occurs . in ( damico et al , 2013 ) , different semi - markov models were applied to wind speed modeling and it was shown that the semi - markov models over perform the markov models and therefore they should be preferred in the modeling of wind speed . in order to better represent the statistical characteristics of wind speed , in a recent article , the idea of an ismc model was advanced in the field of wind speed , see @xcite .
the novelty , with respect to the semi - markov case , consists in the introduction of a third random variable defined as follow : @xmath19 this variable can be interpreted as a moving average of order @xmath20 executed on the series of the past wind speed values @xmath21 with weights given by the fractions of sojourn times in that wind speed @xmath22 with respect to the interval time on which the average is executed @xmath23 .
the ismc model considers that the probability of changes in wind speed do depends also on this new variable . in a simple semi - markov model it depends only on the present wind speed and on the transition time .
the ismc model can be considered as an @xmath24 order semi - markov chain where the dependence is given only by the averaged last @xmath24 states .
also the process @xmath25 has been discretized , table [ um ] shows the states of the process and their values . 1 & 0 to 2.1
+ 2 & 2.1 - 2.6 + 3 & 2.6 - 3.4 + 4 & 3.4 - 6 + 5 & @xmath156 + the parameter @xmath24 must be optimized as a function of the specific database .
the optimization is made by finding the value of @xmath24 that realize the minimum of the root mean square error ( rmse ) between the autocorrelation functions ( acf ) of real and simulated data , see @xcite . in our analysis @xmath26 .
the reasons to introduce this index of memory are found in the presence of a strong autocorrelation that characterize the wind speed process . in the same work
we have shown that if a too small memory is used , the autocorrelation is already persistent but decreases faster than real data . with a longer memory
the autocorrelation remain high for a very long period and also its value is very close to that of real data . if @xmath24 is increased further the autocorrelation drops again to small values .
this behavior suggests the existence of an optimal memory @xmath24 . in our opinion
one can justify this behavior by saying that short memories are not enough to identify in which status ( low , medium low , medium , medium high , high , see table 2 ) is the index @xmath25 , too long memories mix together different status and then much of the information is lost in the average
. the one step transition probability matrix can be evaluated by considering the counting transition between the three random variables considered before .
then , the probability @xmath27 represents the transition probability from the actual wind speed state @xmath28 , to the wind speed state @xmath29 , given that the sojourn time spent in the state @xmath28 is equal to @xmath30 and the value of the process @xmath25 is @xmath31 .
these probabilities can be computed as : @xmath32 [ pri ] where @xmath33 is the total number of transitions observed in the database from state @xmath28 to state @xmath29 in next period having a sojourn time spent in the wind speed @xmath28 equal to @xmath30 and the value of the index process equal to @xmath31 .
the ismc model revealed to be particularly efficient in reproducing together the probability density function of wind speed and the autocorrelation function , see @xcite .
in this section the ismc model is used to forecast future wind speed states by using a one step ahead forecasting procedure , for different time horizons and for various time scales .
particularly , we tested our model using the previously described databases with a sampling frequency of 10 minutes , 30 minutes , 1 hour and 2 hours . for each one of the sampling frequencies ,
the database is divided into two subsets : the first part is used to find the transition probability matrix ( as described in the previous section ) , we will call this part the setting period ; the second part is used to compare the model forecasting with real data ( called testing period ) . as a first attempt to verify the model performance , we used two years of data as setting period and one year as testing . once the transition matrix is set , the forecasted states are computed as follows : @xmath34 where @xmath10 is the number of states in which wind speed is discretized and @xmath27 is the transition probability matrix .
the formula represents the expected value of the next transition given that the present wind speed value is @xmath28 , the sojourn time spent in the state @xmath28 is equal to @xmath30 and the value of the index process @xmath35 is @xmath31 . in figure [ figone ]
we show the results obtained using our model for the four different time scales . in the figure the black continuous line represents real data while the dashed red line is the predicted series . in this figure
the predicted series are long 100 time horizon ( specific time depending on the sampling frequency ) . already from this figure
, it is possible to note that the goodness of the prediction does not fall down at the increasing of the length of the forecasted series .
this is an important result comparing our model with others used in literature for the same purpose .
once the model is set , for a specific site , the forecasting can be made for a long time .
to better verify this point , in the table [ tab1 ] we show quantitative results of our forecasting model for all the considered time horizons .
this table shows the mean values and the standard deviations of the rmse and the mae made on 50 different forecasted series for a certain time scale .
this table shows that the goodness of prediction remains almost constant even varying time scales and time horizons .
.rmse and mae between real wind speed and forecasted series for different time horizon and time scale .
the values are expressed in m / s . [ cols="<,^,^,^,^",options="header " , ]
an important aspect of wind speed model is that they can be used to forecast the economic validity of an investment in wind farm .
we here show how the ismc model can be used with this aim .
we choose to evaluate the internal rate of return , the duration and the convexity of the investment on a wind farm in the chosen specific site .
we conduced two separate analysis : in the first one we consider an investment for a period of 15 years , which corresponds to the period in which the government incentives on the energy price are still present . in the second case
we consider an investment without the government incentives and in this case , being the price of the energy lower than in the previous case , the period of the investment is greater and equal to 30 years . for the analysis we use the database of wind speed described before .
given that it is composed of only 5 years of wind speed we apply a bootstrap procedure in order to have the sufficient number of data for the analysis .
the database has been adapted to the blade height by using the mother described in equation [ height ] . to compare real data with the ismc model we generate , through monte carlo simulation , 1000 possible scenarios of wind speed time series of length 15 years for the firs case and 30 years for the second case ,
the sampling period is 10 minutes . according to the power curve described in figure [ pc1 ]
wind speed has been transformed into power , and then , in produced energy at a specific time window . in figure [ enr1 ] it is shown the comparison between the distribution of the produced energy in a 10 minutes interval for real and synthetic data .
being the transformation from wind speed into energy not linear , this comparison shows that our model is able to reproduce correctly also a real case of energy production .
in fact the values obtained in the two cases are almost identical .
the particular form of these distributions is due to the existence of a cut - in wind speed , under which no energy is produced , and a part of the power curve that remain with constant value , from 10 @xmath36 until 32 @xmath36 , in which the wind turbine produce energy at its rated power .
this explain the existence of the two modes . in figure [ enr2 ]
we show the distribution of energy produced in 1 year .
the comparison is made again between real and synthetic data .
the mean value and the standard deviation of the real annual energy produced are @xmath37 instead for the simulated case we have @xmath38 .
the cut part of the right side of each histogram is due to the presence of a cut - off wind speed , that is the value of the wind speed at which the wind turbine is stopped for structural reason . for wind speed over this value
there is no production of electrical energy .
we will compare some financial indicators computed on the real time series , generated through the bootstrapping procedure explained before , and the simulated date generated through the ismc model by means of monte carlo simulation .
just to remind , we are going to consider two different cases . in the first case
we compute the cash flow obtained by the selling of energy , produced by the wind turbine , taking into account the government incentives on the selling price . in italy for a wind farm with a rated power under 1 @xmath39 and until 15 years from the setup of the plant , the price of the energy is fixed , at the time of writing , to 0.291 @xmath40 . in this case
we consider a cash flow of 15 years .
the second case takes into account the selling of energy at the market price , without the government incentives . in this second case
we fix the energy selling price at 0.027 @xmath40 and , being it smaller than the price with the incentives , the considered period of the cash flow is extended to 30 years . for each case
we generate 1000 possible scenarios and we evaluate the financial indicators for both real and synthetic data . the trajectories are transformed into power through the power curve of the wind turbine chosen , then in energy by considering the time between two successive steps and after in euro by multiplying for the specific price considered . the annual cash flow
are built and at time @xmath41 is considered the coast of the initial investment .
commonly for wind turbines with a rated power under 100 @xmath42 it is considered a price of 3000 for each @xmath42 .
this price takes into account the main cost of the turbine , the setup cost and also the annual maintenance .
being our wind turbine a 10 @xmath42 of rated power , we can easily choose as initial investment 30000 .
at last the financial indicators for all the scenarios are evaluated to compare the results obtained by real data and the synthetic one .
the indicators chosen are the irr ( internal rate of return ) , the duration and the convexity .
the irr is the annualized effective compounded return rate that makes the net present value of all cash flows equal to zero , then it can be evaluated as follow : @xmath43 where @xmath44 is the cash flow of the year @xmath45 , @xmath46 is the internal rate of return and @xmath47 is the number of years of the investment .
the second indicator ( duration ) represents the length of the mean life of the cash flow of an investment weighted with their actualized values and can be formalized as follow : @xmath48 the last financial indicator is the convexity that is a measure of the sensitivity of the duration of an investment to changes in interest rates .
this indicator can be evaluated with the following formula : @xmath49 the notation inside the two previous formulas of the duration and convexity are the same of the irr with the difference of the interest rate @xmath46 that in these two last cases we consider equal to 3% .
figures [ hist1 ] , [ hist2 ] , [ hist3 ] show the histograms of the financial indicators evaluated for all the 1000 scenarios , for real and simulated data , and for the two cases with and without the government incentives .
as it is possible to note , for each pair of indicators and for each specific case of the period of investment , the mean values of the indicators are almost the same for real and simulated data .
in previous works we presented new stochastic models , all based on a semi - markov approach , to generate synthetic time series of wind speed .
we showed that all the models perform better than corresponding markov chain based models in reproducing statistical features of wind speed . using these results , here
, we tried to apply the model which we recognized to be the best among those , namely the indexed semi - markov chain ( ismc ) model , to forecast future wind speed in a specific site .
the ismc model is a nonparametric model and because of this it does not need any assumption on the distribution of wind speed and on wind speed variations .
the results presented in this paper show that the model can be efficiently used to forecast wind speed at different time horizons .
the forecast performance is almost independent from the time horizon used to forecast ; the model can be used without degradation during the considered horizon time , at different time scales ( we showed this for time scales ranging from 10 minutes to 2 hours ) .
the number of data needed to reach a good forecast performance do depend on the time scale used for forecasting ; the model always works better than a simple persistence model .
all these characteristics suggest that the advanced ismc model may be used both for modeling wind speed data and for wind speed prediction .
therefore , it may be utilized as input data for any wind energy system .
h. liua , h. tiana , c. chenb , y. lia . a hybrid statistical method to predict wind speed and wind power .
_ renewable energy _
* 35 * ( 2010 ) 1857 - 1861 + s. salcedo - sanz , a.m. perez - bellido , e.g. ortiz - garcia , a. portilla - figueras , l. prieto , d. paredes .
hybridizing the fifth generation mesoscale model with artificial neural networks for short - term wind speed prediction .
_ renewable energy _
* 34 * ( 2009 ) 1451 - 1457 .
g. damico , f. petroni .
age - usage semi - markov models , _ applied mathematical modelling _ ,
* 35 * ( 2011 ) 4354 - 4366 .
+ g. damico , f. petroni .
weighted - indexed semi - markov models for modeling financial returns , _ journal of statistical mechanics : theory and experiment _ , ( 2012 ) p07015 .
+ s. saurabh , h. zareipour . a review of wind power and wind speed forecasting methods with different time horizons . _ ieee _ north american power symposium ( 2010 ) .
|
the prediction of wind speed is very important when dealing with the production of energy through wind turbines . in this paper , we show a new nonparametric model , based on semi - markov chains , to predict wind speed .
particularly we use an indexed semi - markov model that has been shown to be able to reproduce accurately the statistical behavior of wind speed .
the model is used to forecast , one step ahead , wind speed . in order to check the validity of the model
we show , as indicator of goodness , the root mean square error and mean absolute error between real data and predicted ones . we also compare our forecasting results with those of a persistence model . at last ,
we show an application of the model to predict financial indicators like the internal rate of return , duration and convexity .
|
Subsets and Splits
Data Mining Abstracts
This query searches for and retrieves 2000 documents containing the phrase 'data mining' in the abstract, providing limited but specific document samples.